Expressway emergency rescue early warning system

Abstract

The invention discloses a highway emergency rescue early warning system, which relates to the technical field of intelligent early warning, comprises a division module, a selection module and an early warning module, and is used for dividing severity levels, obtaining resource data, carrying out first selection, checking the resource data, carrying out secondary distribution and carrying out early warning. The accident severity level is quantified in a second level, once the level is locked, a rescue spectrogram gives fixed vehicle and personnel configuration, zero manual intervention of data, level and resources is realized, delay caused by traditional telephone layer-by-layer reporting is avoided, dishes are directly ordered according to orders no matter which road section the accident occurs, the problems of wrong vehicle type adjustment and less equipment are solved, and the service life of the system is prolonged. And the arrival time of the rescue station and the arrival time of the toll station are calculated at the same time, the first selection is made by using the same algorithm, secondary allocation is automatically triggered during missing, supplementary adjustment is carried out from adjacent stations or an emergency linkage library, and level and resource alignment is ensured.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of intelligent early warning, in particular to an expressway emergency rescue early warning system. BACKGROUND

[0002] In recent years, the expressway emergency rescue early warning technology has developed from "passive response" to "active prevention and control", and significant progress has been made in terms of sensing accuracy, early warning timeliness, linkage efficiency and the like, and with the in-depth application of digital twinning, V2X, low-carbon operation and maintenance and the like, the early warning system is more intelligent, accurate and collaborative, thereby providing stronger support for ensuring the safe operation of expressways.

[0003] At present, in the Chinese patent with the publication number CN106627421A, an expressway rescue system and method for electric vehicles with early warning analysis functions are disclosed, the method is sent to a rescue center or a vehicle-mounted system installed on a vehicle through a network; the rescue center or the vehicle-mounted system installed on the vehicle performs scheduling according to the received decision information; the operation center generates a final analysis result according to an abnormal processing model, and completes event storage, early warning analysis and rescue guidance according to the data uploaded by the vehicle-mounted system. The application can analyze the factors that cause faults of expressway electric vehicle charging facilities and affect the charging of expressway electric vehicles, and formulate an expressway electric vehicle emergency rescue scheme, but in the related technology, the starting location is not optimally distributed according to the emergency degree of the rescue task, which is not conducive to the reasonable scheduling of resources, and the rescue distribution is not secondarily scheduled according to the actual remaining resources, which is not conducive to the perfection and comprehensiveness of the rescue. SUMMARY

[0004] The technical problem solved by the application is that in the related technology, the starting location is not optimally distributed according to the emergency degree of the rescue task, which is not conducive to the reasonable scheduling of resources, and the rescue distribution is not secondarily scheduled according to the actual remaining resources, which is not conducive to the perfection and comprehensiveness of the rescue.

[0005] To solve the above technical problems, the application provides the following technical scheme: an expressway emergency rescue early warning system, comprising a division module, a selection module and an early warning module; The division module divides the severity level according to accident-related data, and obtains resource data according to a preset rescue spectrum and the severity level; The selection module obtains a target rescue station according to accident-related data and rescue station distribution, and counts the arrival time length of the target rescue station, obtains a target toll station according to toll station distribution, and counts the arrival time length of the target toll station, and performs first selection on the target rescue station or the target toll station according to the severity level, the arrival time length of the target rescue station and the arrival time length of the target toll station; The pre-warning module verifies the resource data according to the first selection result, and performs secondary distribution and pre-warning according to the verification result.

[0006] As a preferred scheme of the expressway emergency rescue pre-warning system, the division module divides the severity level according to the accident-related data, and obtains the resource data according to a preset rescue spectrum and the severity level. The accident-related data includes vehicle deformation degree, number of accident vehicles, road congestion degree and accident location. The vehicle deformation degree is obtained by extracting shape feature quantity through machine vision and calculating through a cosine similarity formula, and the vehicle deformation degree includes first deformation degree, second deformation degree and third deformation degree, wherein the first deformation degree, the second deformation degree and the third deformation degree represent the deformation degree of the vehicle in increasing order. The severity level includes first level, second level and third level, and is obtained by first analysis. The resource data includes rescue vehicle type, rescue vehicle quantity and rescue personnel quantity, and the rescue vehicle type includes large-scale towing and lifting obstacle rescue vehicle, medium-scale towing and lifting obstacle rescue vehicle, small-scale towing and lifting obstacle rescue vehicle, medium-scale truck, small-scale truck and patrol pickup truck. The preset rescue spectrum includes configuration logic of resource data corresponding to accidents of different severity levels.

[0007] As a preferred scheme of the expressway emergency rescue pre-warning system, the calculation logic of the vehicle deformation degree includes: Obtain a picture of any accident vehicle, determine the vehicle type of the accident vehicle, the vehicle type includes small vehicle, medium vehicle and large vehicle, call the standard deformation degree picture corresponding to the vehicle type of the accident vehicle, extract the first feature quantity of each standard deformation degree picture, and extract the second feature quantity of the picture of the accident vehicle, the first feature quantity and the second feature quantity are represented as shape feature quantity, the first similarity of the first feature quantity and the second feature quantity is calculated through a cosine similarity formula, the first value is set as the first similarity threshold, the first similarity is compared with the first value, when the first similarity is less than the first value, the next first similarity is jumped, when the first similarity is greater than or equal to the first value, the standard deformation degree picture corresponding to the first similarity is set as the deformation picture of the accident vehicle, and the corresponding vehicle deformation degree is obtained.

[0008] As a preferred scheme of the expressway emergency rescue pre-warning system, the logic of obtaining the severity level through first analysis includes: Obtaining the vehicle deformation degree, the number of accident vehicles and the road congestion degree, calculating a first sum of values of the vehicle deformation degree, the number of accident vehicles and the road congestion degree, setting a second value and a third value as a first sum value threshold, wherein the numerical size order of the second value and the third value is from small to large, comparing the first sum value with the first sum value threshold, when the first sum value is less than or equal to the second value, setting the severity level as a first level, when the first sum value is greater than the second value and less than or equal to the third value, setting the severity level as a second level, and when the first sum value is greater than the third value, setting the severity level as a third level.

[0009] As a preferred scheme of the expressway emergency rescue early warning system, the logic of obtaining the resource data according to the preset rescue spectrum and the severity level comprises: A two-dimensional table is used to give the mapping relationship between the severity level and the vehicle type, the vehicle number and the personnel number. The first sum value is obtained in real time, and the corresponding vehicle type, vehicle number and personnel number are found according to the rescue spectrum.

[0010] As a preferred scheme of the expressway emergency rescue early warning system, when the severity level is the first level, the mapping output is one small towing and obstacle removal rescue vehicle, one patrol pickup truck and three personnel. When the severity level is the second level, the mapping output is one medium towing and obstacle removal rescue vehicle, one small towing and obstacle removal rescue vehicle, one small truck, two patrol pickup trucks and seven personnel. When the severity level is the third level, the mapping output is one large towing and obstacle removal rescue vehicle, one medium towing and obstacle removal rescue vehicle, two small towing and obstacle removal rescue vehicles, one medium truck, one small truck, three patrol pickup trucks and fifteen personnel.

[0011] As a preferred scheme of the expressway emergency rescue early warning system, the logic of obtaining the target rescue station according to the accident related data and the rescue station distribution comprises: The accident position is obtained, the position is represented by the accuracy and the latitude, the positions of the rescue stations are obtained, the first distances between the positions of the rescue stations and the accident position are counted, the first distances are sorted in ascending order, the first distances with the top two values are selected, the corresponding rescue stations are obtained and are recorded as the target rescue stations. The logic of obtaining the target toll station according to the toll station distribution comprises: The positions of the toll stations are obtained, the second distances between the positions of the toll stations and the accident position are counted, the second distances are sorted in ascending order, the second distances with the top two values are selected, the corresponding toll stations are obtained and are recorded as the target toll stations.

[0012] As a preferred scheme of the expressway emergency rescue early warning system, the historical rescue data includes historical rescue starting locations, historical rescue distances and historical rescue time lengths, the historical rescue starting locations include rescue stations and toll stations, the historical rescue distances are expressed as distances from the toll stations or the rescue stations to the accident locations, when the historical rescue starting location is a toll station, the corresponding historical rescue time length includes a turning time length, a passing time length and a reverse time length, and when the historical rescue starting location is a rescue station, the corresponding historical rescue time length is expressed as a passing time length.

[0013] As a preferred scheme of the expressway emergency rescue early warning system, the calculation method of the target toll station arrival time length includes: When the historical rescue starting location is a toll station, a second sum value of the historical rescue time lengths is calculated, a first ratio value of each second sum value and the corresponding historical rescue distance is calculated by traversing each second sum value, a first average value of the first ratio values is calculated by traversing each first ratio value, and the first average value is expressed as a comprehensive time length required for each kilometer when starting from the toll station; A first product of the second distance and the first average value is calculated, the first product is set as the target toll station arrival time length, a toll station corresponding to a target toll station arrival time length with a smaller value is set as the first toll station, and a toll station corresponding to a target toll station arrival time length with a larger value is set as the second toll station; The calculation method of the target rescue station arrival time length includes: When the historical rescue starting location is a rescue station, a second ratio value of the historical rescue time length and the corresponding historical rescue distance is calculated, a second average value of the second ratio values is calculated by traversing each second ratio value, and the second average value is expressed as a time length required for each kilometer when starting from the rescue station; A second product of the first distance and the second average value is calculated, the second product is set as the target rescue station arrival time length, a rescue station corresponding to a target rescue station arrival time length with a larger value is set as the first rescue station, and a rescue station corresponding to a target rescue station arrival time length with a larger value is set as the second rescue station; The logic of the first selection of the target rescue station or the target toll station according to the severity level, the target rescue station arrival time length and the target toll station arrival time length includes: The severity level is obtained, when the severity level is a first level, a toll station corresponding to a smaller value or a larger value in the target toll station arrival time length is selected, and the first selection result is set as the first toll station or the second toll station; When the severity level is a second level, a rescue station corresponding to a larger value in the target rescue station arrival time length is selected, and the first selection result is set as the first rescue station; When the severity level is the third level, the target rescue station is selected according to the smaller value of the rescue station arrival time length, and the first selection result is set as the second rescue station.

[0014] As a preferred scheme of the expressway emergency rescue pre-warning system, the method for verifying the resource data according to the first selection result comprises: The first selection result is obtained, and the residual resource data of the toll station or rescue station corresponding to the first selection result is obtained, and the residual resource data is compared with the resource data, when the residual resource data is less than the resource data, secondary allocation is performed; When all the residual resource data is greater than or equal to the resource data, no secondary allocation is performed; The logic of the secondary allocation comprises: The first distance of the top 4 corresponding toll stations or rescue stations is selected, and a secondary allocation request is sent to the residual toll station or rescue station of the same type as the first selection result, and the request content represents the resource data lacking at the first selection result; The method for pre-warning according to the verification result comprises: When the secondary allocation is needed, the first pre-warning is performed, and when the secondary allocation is not needed, the first pre-warning is not performed.

[0015] The beneficial effects of the present application are as follows: the accident severity level is quantified by seconds, once the level is locked, the rescue spectrum gives a fixed vehicle and personnel configuration, the data, level and resource are realized without manual intervention, the delay caused by the traditional telephone layer-by-layer reporting is avoided, no matter which road section the accident occurs, the single-point menu is directly selected, the problems of previous vehicle type adjustment and less equipment are solved, the secondary vehicle dispatching time is reduced, the rescue station arrival time length and the toll station arrival time length are calculated, the same algorithm is used for the first selection, for the high-risk road sections such as tunnel groups and long downhills, the toll station can be used as a front rescue point, when the composition is insufficient, the secondary allocation is automatically triggered, the adjacent stations or emergency linkage warehouses are adjusted, and the level and resource are aligned. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 A basic flowchart of an expressway emergency rescue pre-warning system is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0017] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application are described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0018] Embodiment, refer to Figure 1For an embodiment of the present application, a highway emergency rescue early warning system is provided, comprising a division module, a selection module and a warning module; The division module divides the severity level according to the accident-related data, and obtains resource data according to the preset rescue spectrum and the severity level; The selection module obtains the target rescue station according to the accident-related data and the distribution of rescue stations, and counts the arrival time of the target rescue station, obtains the target toll station according to the distribution of toll stations, and counts the arrival time of the target toll station, and makes a first selection of the target rescue station or the target toll station according to the severity level, the arrival time of the target rescue station and the arrival time of the target toll station; The warning module verifies the resource data according to the first selection result, and makes a second distribution and warning according to the verification result.

[0019] The present application quantifies the severity level of the accident in seconds, and once the level is locked, the rescue spectrum gives a fixed vehicle and personnel configuration, realizing zero manual intervention of data, level and resources, avoiding the delay caused by traditional telephone layer-by-layer reporting, and directly selecting a single point menu regardless of the road section where the accident occurs, solving the problem of incorrect vehicle type and insufficient equipment in the past, and reducing the time of secondary vehicle dispatching, while calculating the arrival time of the rescue station and the arrival time of the toll station, using the same algorithm for first selection, for high-risk road sections such as tunnel groups and long downhills, the toll station can be used as a front rescue point, and when the train is short, the second distribution is automatically triggered, and the train is supplemented from the adjacent station or emergency linkage warehouse, to ensure that the level and resources are aligned.

[0020] The division module divides the severity level according to the accident-related data, and obtains resource data according to the preset rescue spectrum and the severity level; The accident-related data includes vehicle deformation degree, number of accident vehicles, road congestion degree and accident location; The vehicle deformation degree is obtained by extracting shape feature quantity through machine vision and calculating through cosine similarity formula, and the vehicle deformation degree includes first deformation degree, second deformation degree and third deformation degree, wherein the first deformation degree, the second deformation degree and the third deformation degree represent the deformation degree of the vehicle is getting larger and larger; The severity level includes first level, second level and third level, which is obtained by first analysis; The resource data includes rescue vehicle type, rescue vehicle quantity and rescue personnel quantity, and the rescue vehicle type includes large-scale towing and lifting obstacle rescue vehicle, medium-sized towing and lifting obstacle rescue vehicle, small-sized towing and lifting obstacle rescue vehicle, medium-sized truck, small-sized truck and patrol pickup truck; The preset rescue spectrum includes the configuration logic of resource data corresponding to accidents of different severity levels.

[0021] In a specific implementation, the first / second / third deformation degree increases in degree of deformation (slight to moderate to severe), and is one-to-one corresponding to the subsequent three grades, avoiding the error of traditional "eyeball damage estimation" of ±30%. The cosine similarity is completed in a 512-dimensional shape feature space, and the single picture inference time is 42ms (NVIDIA Jetson Xavier measured). Under 4G network, the deformation degree value can be returned within 1.2s, realizing "shooting and calculating at the same time". The camera is 4K@30fps, FOV110°, the vertical rod height is 8m, covering 3 lanes, the edge computing power is Jetson Xavier, FP16 inference, power consumption is 30W, the dataset is 110,000 accident vehicle pictures, mIoU0.93, deformation degree three-grade classification accuracy is 95.7%, the first-grade accident: 36% is changed by the toll station to be disposed by the light-duty tow truck first, saving 8.4km of empty running, the third-grade accident: 100% is still handled by the large-scale towing and lifting of the rescue station, ensuring that heavy equipment is in place, 5G SA network, end-to-end average latency is 19ms, Beidou+RTK positioning accuracy is ≤20cm, ensuring that the stake number error is <5m. Compared with the traditional expansion rescue station scheme, the civil engineering and equipment investment is saved by 72%.

[0022] The calculation logic of the vehicle deformation degree includes: An image of any accident vehicle is obtained, the vehicle type of the accident vehicle is judged, the vehicle type includes a small vehicle, a medium vehicle and a large vehicle, the standard deformation degree picture corresponding to the vehicle type of the accident vehicle is called, the first feature quantity of each standard deformation degree picture is extracted, and the second feature quantity of the picture of the accident vehicle is extracted. The first feature quantity and the second feature quantity are both expressed as shape feature quantities, the first similarity of the first feature quantity and the second feature quantity is calculated by a cosine similarity formula, the first value is set as a first similarity threshold, the first similarity is compared with the first value, when the first similarity is less than the first value, the next first similarity is jumped, when the first similarity is greater than or equal to the first value, the standard deformation degree picture corresponding to the first similarity is set as the deformation picture of the accident vehicle, and the corresponding vehicle deformation degree is obtained.

[0023] In specific implementation, the three sets of template libraries for small cars, medium cars and large cars are independent of each other, and only cosine search needs to be performed in the corresponding sub-library. The number of comparisons of a single picture is reduced from 3xN to N. The actual inference time on Jetson Xavier is 42 ms→17 ms. The first value (cosθ≥0.85) is used as the "sufficiently similar" gate. If the value is below the threshold, it is directly skipped. In the field, 87 real accidents have been verified for 6 months. No grade misjudgment caused by template mismatching has occurred. The camera takes pictures→transmits to the cloud through 5G→GPU inference→returns the deformation degree. The complete link takes an average of 1.8 seconds, which meets the system index of "completing grade division in 1 minute". Only 3 standard deformation degree pictures (mild, moderate and severe) are retained for each type of vehicle. The compressed single picture is 120 kB. Compared with the traditional 3000 full-quantity picture library, the storage is reduced by 98%, and the monthly traffic is saved by 46 GB. The template pictures are collected once and written to the edge node. Subsequent addition of new vehicle types only needs to upload 3 new templates to automatically expand the capacity without manual annotation of new data sets.

[0024] The logic of obtaining the severity level through the first analysis includes: The vehicle deformation degree, the number of accident vehicles and the road congestion degree are obtained. The first sum of the numerical values of the vehicle deformation degree, the number of accident vehicles and the road congestion degree is calculated. The second value and the third value are set as the first sum threshold. The numerical values of the second value and the third value are in the order of small to large. The first sum is compared with the first sum threshold. When the first sum is less than or equal to the second value, the severity level is set as the first level. When the first sum is greater than the second value and less than or equal to the third value, the severity level is set as the second level. When the first sum is greater than the third value, the severity level is set as the third level.

[0025] In specific implementation, the vehicle deformation degree (cm), the number of accident vehicles (vehicles) and the road congestion degree (0-1 interval) are directly added. The CPU only needs to perform a floating-point addition, which takes an average of 0.05 ms. Compared with the traditional multi-factor weighted model (≥3 times of multiplication + 2 times of addition), the calculation amount is reduced by 90%. The edge MCU can also bear the load. The second value (mild / moderate boundary) and the third value (moderate / severe boundary) are calibrated once during the deployment stage and do not need to be adjusted thereafter. The threshold drift is less than ±2% in the verification of 87 accidents for 6 months. The grade misjudgment rate is only 1.1%. The artificial expert grading is used as the true value. The ROC curve determines the optimal cut-off. The second value is 30 (sensitivity 0.94, specificity 0.91). The third value is 70 (sensitivity 0.96, specificity 0.93). If any sensor is offline, the system defaults the variable to 0, which still guarantees uninterrupted grading. The jumping data uses 3-frame median filtering. The misjudgment rate is less than 0.5% after filtering.

[0026] The logic of obtaining the resource data according to the preset rescue spectrum and the severity level includes: A two-dimensional table is used to give the mapping relationship of the severity level and the vehicle type, the number of vehicles, and the number of personnel; The first sum is obtained in real time, and the corresponding vehicle type, the number of vehicles, and the number of personnel are obtained according to the rescue spectrum.

[0027] The preset rescue spectrum includes: when the severity level is the first level, the mapped output is 1 small towing and rescue vehicle, 1 patrol pickup, and 3 personnel; When the severity level is the second level, the mapped output is 1 medium towing and rescue vehicle, 1 small towing and rescue vehicle, 1 small truck, 2 patrol pickups, and 7 personnel; When the severity level is the third level, the mapped output is 1 large towing and rescue vehicle, 1 medium towing and rescue vehicle, 2 small towing and rescue vehicles, 1 medium truck, 1 small truck, 3 patrol pickups, and 15 personnel.

[0028] In a specific implementation, a 3x4 two-dimensional table is used to directly convert the “severity level” into a complete list of “vehicle type + number + personnel”, and the on-site MCU only needs to perform a table lookup (O(1) complexity) once, with an average instruction generation time of 0.9 ms, which is 200 times faster than the traditional rule engine. Each level has a fixed combination, which eliminates the secondary dispatch caused by “fewer vehicles” and “wrongly dispatched vehicles”. In 6 months of on-site accident verification, the vehicle type was 100% accurate, and the personnel were 100% accurate. Compared with manual experience scheduling, the empty mileage is reduced by 28%, the fuel is saved by 21%, the inventory turnover rate of the rescue station / toll station is increased by 35%, the long-term idle equipment is reduced, the dispatcher only needs to confirm the “level->send” two actions, and the traditional telephone check of vehicle type and number is cancelled. 4-6 links, an average of 1.5 min is saved.

[0029] According to the accident-related data and the distribution of the rescue station, the logic of the target rescue station includes: The accident location is obtained, the location is represented by the precision and the latitude, the location of each rescue station is obtained, the first distance between the location of the rescue station and the accident location is counted, the first distance is sorted in ascending order, the first distance with the top 2 values is selected, and the corresponding rescue station is obtained, which is recorded as the target rescue station; According to the distribution of the toll station, the logic of the target toll station includes: The location of each toll station is obtained, and the second distance between the location of the toll station and the accident location is counted, the second distance is sorted in ascending order, the second distance with the top 2 values is selected, and the corresponding toll station is obtained, which is recorded as the target toll station.

[0030] obtaining historical rescue data, the historical rescue data including historical rescue starting locations, historical rescue distances and historical rescue time lengths, the historical rescue starting locations including rescue stations and toll stations, the historical rescue distances being represented as distances from the toll stations or the rescue stations to accident locations, when the historical rescue starting location is a toll station, the corresponding historical rescue time length including a turning time length, a passing time length and a reverse time length, and when the historical rescue starting location is a rescue station, the corresponding historical rescue time length being represented as the passing time length.

[0031] In the specific implementation, the two rescue stations closest in distance and the two toll stations closest in distance are both pulled into the candidate pool, and the selectable range is expanded by 3.8 times. In a 118 km demonstration section, the average distance between toll stations is 18 km, and the average distance between rescue stations is 42 km. The double-source strategy increases the road section that can be covered within 10 minutes from 64% to 93%. When the starting location is a toll station, the turning time length, the passing time length and the reverse time length have been disassembled from the historical records, and the prediction error is less than or equal to ±8%. When the starting location is a rescue station, only the passing time length is retained, and the prediction error is less than or equal to ±5%. The true time length is used to replace the pure distance / speed estimation, and the dispatching accuracy is increased from 72% to 96%. Only the existing Beidou trajectory and ETC gantry data are read, and no additional roadside sensors are needed. The algorithm runs entirely on the existing cloud control platform, and the CPU overhead is less than 1%.

[0032] The calculation method of the target toll station arrival time length includes: When the historical rescue starting location is a toll station, a second sum value of the historical rescue time lengths is calculated, each second sum value is traversed, a first ratio value of the second sum value and the corresponding historical rescue distance is calculated, each first ratio value is traversed, a first average value of the first ratio values is calculated, and the first average value is represented as a comprehensive time length required per kilometer when starting from the toll station; A first product of the second distance and the first average value is calculated, the first product is set as the target toll station arrival time length, a toll station corresponding to a target toll station arrival time length with a smaller value is set as the first toll station, and a toll station corresponding to a target toll station arrival time length with a larger value is set as the second toll station; The calculation method of the target rescue station arrival time length includes: When the historical rescue starting location is a rescue station, a second ratio value of the historical rescue time length and the corresponding historical rescue distance is calculated, each second ratio value is traversed, a second average value of the second ratio values is calculated, and the second average value is represented as a time length required per kilometer when starting from the rescue station; A second product of the first distance and the second average value is calculated, the second product is set as the target rescue station arrival time length, a rescue station corresponding to a target rescue station arrival time length with a larger value is set as the first rescue station, and a rescue station corresponding to a target rescue station arrival time length with a larger value is set as the second rescue station; The logic of the first selection of the target rescue station or the target toll station according to the severity level, the target rescue station arrival time and the target toll station arrival time comprises: Obtaining the severity level, when the severity level is the first level, selecting the toll station corresponding to the smaller value or the larger value in the target toll station arrival time, and setting the first selection result as the first toll station or the second toll station; When the severity level is the second level, selecting the rescue station corresponding to the larger value in the target rescue station arrival time, and setting the first selection result as the first rescue station; When the severity level is the third level, selecting the rescue station corresponding to the smaller value in the target rescue station arrival time, and setting the first selection result as the second rescue station.

[0033] In the specific implementation, the first average value (min / km) is generated by the second sum of the three historical time lengths of “turning + passing + reversing” for the toll station model, the prediction error is only ±0.4 min, the second average value (min / km) is generated by the single passing time length for the rescue station model, and the error is ±0.3 min, the toll station: taking the minimum arrival time → the first toll station, the second minimum arrival time → the second toll station, the rescue station: taking the minimum arrival time → the second rescue station, the second minimum arrival time → the first rescue station (opposite to the toll station number, which is convenient for unified comparison of downstream logic), the first level → selecting the fastest toll station (small accident, light equipment can be used), the second level → selecting the slowest rescue station (medium-sized accident, medium-sized towing is needed, and more waiting time of 2-3 min is preferred to ensure the equipment), and the third level → selecting the fastest rescue station (large accident, heavy towing is needed, and every second counts).

[0034] The method for verifying the resource data according to the first selection result by the early warning module comprises: Obtaining the first selection result, obtaining the remaining resource data of the toll station or the rescue station corresponding to the first selection result, comparing the remaining resource data with the resource data, and when the remaining resource data is less than the resource data, performing secondary distribution; When all the remaining resource data is greater than or equal to the resource data, no secondary distribution is performed; The logic of the secondary distribution comprises: Selecting the corresponding toll station or rescue station in the top 4 of the first distance, and sending a secondary distribution request to the remaining toll station or rescue station of the same type as the first selection result, and the request content indicates the resource data lacking at the first selection result; The method for early warning according to the verification result comprises: When the secondary distribution is needed, the first early warning is performed, and when the secondary distribution is not needed, the first early warning is not performed.

[0035] In a specific implementation, before the rescue instruction is issued, a one-time check is performed using "remaining resources > required resources" to avoid the traditional scenario of "returning empty after arriving at the scene without a car"; in 87 accidents in 6 months, the secondary empty return rate decreased from 12% to 0, the same site shared car models, reduced cross-model scheduling, and the success rate of position filling was 100%, the need for secondary allocation triggered "first warning" (SMS + voice + large screen red), the command center received it within 0.3s, and did not need secondary allocation to pass silently and avoid information noise.

[0036] The application quantifies the severity level of the accident in seconds, and once the level is locked, the rescue spectrum gives a fixed vehicle and personnel configuration, realizing zero manual intervention of data, level and resources, avoiding the delay caused by traditional telephone reporting, and directly selecting a single point menu regardless of the road section where the accident occurs, solving the problem of incorrect car type and insufficient equipment in the past, and reducing the time of secondary vehicle dispatching. At the same time, the arrival time of the rescue station and the arrival time of the toll station are calculated, and the same algorithm is used for the first selection. For high-risk road sections such as tunnel groups and long downhills, the toll station can be used as a front rescue point, and when there is a lack of coding, secondary allocation is automatically triggered to supplement from adjacent stations or emergency linkage warehouses to ensure that the level and resources are aligned.

[0037] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems or computer program products. Therefore, the application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Moreover, the application can take the form of a computer program product implemented on one or more computer-usable storage media having computer-usable program code embodied in the medium. The storage media can be realized by any type of volatile or non-volatile storage devices or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. These computer program instructions can also be stored in a computer readable storage medium that can guide a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer readable storage medium produce a product including instruction means, which realizes the flow Figure 1 one flow or multiple flows and / or blocks Figure 1the function specified in the one or more blocks.

[0038] It should be noted that the above-mentioned embodiments are only used to illustrate the technical solutions of the present application, not to limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalent replaced without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A highway emergency rescue pre-warning system, characterized in that, The method comprises a division module, a selection module and a warning module. The division module divides the severity level according to the accident-related data, and obtains resource data according to the preset rescue spectrum and the severity level. The selection module obtains a target rescue station according to the accident-related data and the distribution of rescue stations, and counts the arrival time of the target rescue station; obtains a target toll station according to the distribution of toll stations, and counts the arrival time of the target toll station; and makes a first selection of the target rescue station or the target toll station according to the severity level, the arrival time of the target rescue station and the arrival time of the target toll station. The warning module verifies the resource data according to the first selection result, and makes a second distribution and gives a warning according to the verification result.

2. The expressway emergency rescue pre-warning system according to claim 1, characterized in that: The division module divides the severity level according to the accident-related data, and obtains resource data according to the preset rescue spectrum and the severity level. The accident-related data includes vehicle deformation degree, number of accident vehicles, road congestion degree and accident location. The vehicle deformation degree is obtained by extracting shape feature quantity through machine vision and calculating by a cosine similarity formula, and the vehicle deformation degree includes first deformation degree, second deformation degree and third deformation degree, wherein the first deformation degree, the second deformation degree and the third deformation degree represent the deformation degree of the vehicle in increasing order. The severity level includes first level, second level and third level, which is obtained by first analysis. The resource data includes rescue vehicle type, rescue vehicle quantity and rescue personnel quantity, and the rescue vehicle type includes large-scale towing and lifting rescue vehicle, medium-scale towing and lifting rescue vehicle, small-scale towing and lifting rescue vehicle, medium-scale truck, small-scale truck and patrol pickup truck. The preset rescue spectrum includes the configuration logic of resource data corresponding to accidents of different severity levels.

3. The expressway emergency rescue pre-warning system according to claim 2, characterized in that: The calculation logic of the vehicle deformation degree includes: Obtain a picture of any accident vehicle, determine the vehicle type of the accident vehicle, the vehicle type includes small car, medium car and large car, call the standard deformation degree picture corresponding to the vehicle type of the accident vehicle, extract the first feature quantity of each standard deformation degree picture, and extract the second feature quantity of the picture of the accident vehicle, the first feature quantity and the second feature quantity are represented as shape feature quantity, the first similarity of the first feature quantity and the second feature quantity is calculated by a cosine similarity formula, the first value is set as the first similarity threshold, the first similarity is compared with the first value, when the first similarity is less than the first value, the next first similarity is jumped, when the first similarity is greater than or equal to the first value, the standard deformation degree picture corresponding to the first similarity is set as the deformation picture of the accident vehicle, and the corresponding vehicle deformation degree is obtained.

4. The expressway emergency rescue pre-warning system according to claim 2, characterized in that: The logic of obtaining the severity level by first analysis includes: The vehicle deformation degree, the number of accident vehicles and the road congestion degree are acquired, a first sum value of the vehicle deformation degree, the number of accident vehicles and the road congestion degree is calculated, a second value and a third value are set as a first sum value threshold, the second value and the third value are in order from small to large, the first sum value is compared with the first sum value threshold, when the first sum value is less than or equal to the second value, the severity level is set as a first level, when the first sum value is greater than the second value and less than or equal to the third value, the severity level is set as a second level, and when the first sum value is greater than the third value, the severity level is set as a third level.

5. The expressway emergency rescue pre-warning system according to claim 4, characterized in that: The logic of obtaining resource data according to the preset rescue spectrum and the severity level comprises: a two-dimensional table is used to give a mapping relationship between the severity level and the vehicle type, the number of vehicles and the number of personnel; the first sum value is acquired in real time, and the corresponding vehicle type, the number of vehicles and the number of personnel are found out according to the rescue spectrum.

6. The expressway emergency rescue pre-warning system according to claim 5, characterized in that: The preset rescue spectrum comprises: when the severity level is the first level, the mapping output is one small towing and obstacle rescue truck, one patrol pickup truck and three personnel; when the severity level is the second level, the mapping output is one medium towing and obstacle rescue truck, one small towing and obstacle rescue truck, one small truck, two patrol pickup trucks and seven personnel; 7. The expressway emergency rescue pre-warning system according to claim 1, wherein: when the severity level is the third level, the mapping output is one large towing and obstacle rescue truck, one medium towing and obstacle rescue truck, two small towing and obstacle rescue trucks, one medium truck, one small truck, three patrol pickup trucks and fifteen personnel. The logic of obtaining the target rescue station according to the accident related data and the distribution of rescue stations comprises: the position of the accident is acquired, the position is expressed as the accuracy and the latitude, the positions of the rescue stations are acquired, the first distances between the positions of the rescue stations and the position of the accident are counted, the first distances are sorted in ascending order, the first distances with the top two values are selected, the corresponding rescue stations are acquired and recorded as the target rescue stations; The logic of obtaining the target toll station according to the distribution of toll stations comprises:

8. The expressway emergency rescue pre-warning system according to claim 1, wherein: the positions of the toll stations are acquired, the second distances between the positions of the toll stations and the position of the accident are counted, the second distances are sorted in ascending order, the second distances with the top two values are selected, the corresponding toll stations are acquired and recorded as the target toll stations.

9. The expressway emergency rescue pre-warning system according to claim 7, characterized in that: The historical rescue data are acquired, the historical rescue data comprise historical rescue starting locations, historical rescue distances and historical rescue time lengths, the historical rescue starting locations comprise rescue stations and toll stations, the historical rescue distance represents the distance from the toll station or the rescue station to the accident site, when the historical rescue starting location is the toll station, the corresponding historical rescue time length comprises the turning time length, the passing time length and the reverse time length, and when the historical rescue starting location is the rescue station, the corresponding historical rescue time length represents the passing time length. The calculation method of the target toll station arrival time length comprises: When the historical rescue starting point is a toll station, a second sum of historical rescue time lengths is calculated, each second sum is traversed, a first ratio of the second sum to a corresponding historical rescue distance is calculated, each first ratio is traversed, a first average of the first ratios is calculated, and the first average represents a comprehensive time length consumed per kilometer from the toll station; a first product of the second distance and the first average is calculated, the first product is set as a target toll station arrival time length, a toll station corresponding to a smaller target toll station arrival time length is set as a first toll station, and a toll station corresponding to a larger target toll station arrival time length is set as a second toll station; a target rescue station arrival time length is calculated by the following method: When the historical rescue starting point is a rescue station, a second ratio of a historical rescue time length to a corresponding historical rescue distance is calculated, each second ratio is traversed, a second average of the second ratios is calculated, and the second average represents a time length consumed per kilometer from the rescue station; a second product of the first distance and the second average is calculated, the second product is set as a target rescue station arrival time length, a rescue station corresponding to a larger target rescue station arrival time length is set as a first rescue station, and a rescue station corresponding to a larger target rescue station arrival time length is set as a second rescue station; the logic of first selecting a target rescue station or a target toll station according to a severity level, a target rescue station arrival time length, and a target toll station arrival time length includes: the severity level is obtained, when the severity level is a first level, a toll station corresponding to a smaller or larger value in the target toll station arrival time length is selected, and a first selection result is set as the first toll station or the second toll station; when the severity level is a second level, a rescue station corresponding to a larger value in the target rescue station arrival time length is selected, and the first selection result is set as the first rescue station; when the severity level is a third level, a rescue station corresponding to a smaller value in the target rescue station arrival time length is selected, and the first selection result is set as the second rescue station.

10. The expressway emergency rescue pre-warning system according to claim 1, wherein: the method of verifying resource data according to the first selection result includes: the first selection result is obtained, and remaining resource data of a toll station or a rescue station corresponding to the first selection result is obtained, the remaining resource data is compared with the resource data, when the remaining resource data is less than the resource data, secondary allocation is performed; when all the remaining resource data is greater than or equal to the resource data, secondary allocation is not performed; the logic of secondary allocation includes: a toll station or a rescue station corresponding to the first distance in the top four is selected, and a secondary allocation request is sent to a remaining toll station or rescue station of the same type as the first selection result, and the request content represents resource data lacking at the first selection result; the method of performing early warning according to the verification result includes: when secondary allocation is needed, first early warning is performed, and when secondary allocation is not needed, first early warning is not performed.

Images