Smart city tunnel traffic emergency supervision internet of things large model system and method

Abstract

The application relates to the field of traffic supervision, and provides a smart city tunnel traffic emergency supervision Internet of Things large model system and method, the method comprising the following steps: constructing a fusion feature vector of a vehicle in a target tunnel according to multidimensional data of the target tunnel; determining an abnormal event according to the fusion feature vector and congestion data of the target tunnel; predicting an upstream congestion condition of the target tunnel according to the abnormal event, the congestion data and surrounding traffic flow data of the target tunnel; and generating a signal control instruction comprising a passing time length according to the upstream congestion condition, the signal control instruction being configured to control a target signal lamp to maintain a target state for the passing time length, the target signal lamp being a signal lamp on a key road section in a direction leading to the target tunnel. The application can realize a leap from passive control of 'getting stuck at the source' to active dredging of 'avoiding the congestion point', provide clear and feasible alternative solutions for drivers, and significantly improve the overall passing efficiency and emergency disposal capability of a regional road network.

Description

Technical Field

[0001] This manual relates to the field of traffic supervision, and in particular to a large-scale IoT model system and method for emergency traffic supervision in smart city tunnels. Background Technology

[0002] As urban populations gradually increase, urban tunnel transportation is also becoming increasingly developed. The regulation of urban traffic has become an indispensable part of improving residents' quality of life.

[0003] The current traffic monitoring system has limited means of sensing traffic incidents, a passive response, and control measures that are limited to local road sections, making it difficult to proactively prevent and fundamentally alleviate congestion.

[0004] To address the aforementioned issues, there is an urgent need for a smart city tunnel traffic emergency monitoring IoT big data model system and method that can link regional road networks to predict surrounding congestion and coordinate traffic conditions. Summary of the Invention

[0005] This specification provides one or more embodiments of a smart city tunnel traffic emergency monitoring IoT big data model system, including an emergency monitoring management platform configured to execute a smart city tunnel traffic emergency monitoring method.

[0006] This specification provides one or more embodiments of a smart city tunnel traffic emergency monitoring method, executed by an emergency monitoring management platform. The method includes: constructing a fusion feature vector of vehicles within the target tunnel based on multidimensional data of the target tunnel; identifying abnormal events based on the fusion feature vector and congestion data of the target tunnel; predicting upstream congestion conditions of the target tunnel based on the abnormal events, the congestion data, and surrounding traffic flow data of the target tunnel; generating a signal control instruction including a passage duration based on the upstream congestion conditions; and controlling a target traffic light to maintain a target state for the passage duration according to the signal control instruction, wherein the target traffic light is a traffic light on a key road section heading towards the target tunnel.

[0007] One or more embodiments of the present invention have at least the following technical effects: (1) It can accurately identify minor abnormal events in the target tunnel and predict the trend and scope of congestion spreading to the upstream road section in advance; it can also control the traffic lights of key roads and actively restrict the merging traffic before serious congestion occurs, thereby realizing the transformation from "passive response to congestion" to "active prevention of congestion", shortening the reaction time from the occurrence of the event to the implementation of diversion; and it can avoid the delay and misjudgment of manual intervention, thereby reducing the risk of regional traffic paralysis caused by sudden events in the target tunnel; (2) On the basis of predicting congestion and controlling traffic lights, it further combines real-time road conditions to actively generate detour routes and release guidance information through various channels such as information boards and vehicle terminals, realizing the leap from passive control of "congestion at the source" to active diversion of "bypassing the congestion point", providing drivers with clear and feasible alternatives, and significantly improving the overall traffic efficiency and emergency response capabilities of the regional road network. Attached Figure Description

[0008] This specification will be further described by way of exemplary embodiments, which will be described in detail with reference to the accompanying drawings. These embodiments are not limiting; in these embodiments, the same reference numerals denote the same structures, wherein:

[0009] Figure 1 This is an exemplary structural diagram of a large-scale IoT model system for smart city tunnel traffic emergency monitoring, as shown in some embodiments of this specification. Figure 2 This is an exemplary flowchart of a smart city tunnel traffic emergency monitoring method according to some embodiments of this specification; Figure 3 These are exemplary schematic diagrams illustrating the determination of passage duration according to some embodiments of this specification; Figure 4 These are exemplary schematic diagrams of anomaly recognition models according to some embodiments of this specification; Figure 5 This is an exemplary schematic diagram illustrating the determination of upstream congestion according to some embodiments of this specification. Detailed Implementation

[0010] To more clearly illustrate the technical solutions of the embodiments in this specification, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are merely some examples or embodiments of this specification. For those skilled in the art, these drawings can be applied to other similar scenarios without creative effort. Unless obvious from the context or otherwise specified, the same reference numerals in the drawings represent the same structures or operations.

[0011] Figure 1This is an exemplary structural diagram of a large-scale IoT model system for smart city tunnel traffic emergency monitoring, as shown in some embodiments of this specification.

[0012] In some embodiments, such as Figure 1 As shown, the smart city tunnel traffic emergency monitoring IoT big data model system 100 (hereinafter referred to as the system) may include an emergency monitoring and management platform 130.

[0013] The Emergency Monitoring and Management Platform 130 refers to a digital monitoring and management platform for supervising traffic emergency events. Traffic emergency events are sudden, destructive, and harmful traffic incidents, such as abnormal events. For an explanation of abnormal events, please refer to [link to relevant documentation]. Figure 2 And its contents.

[0014] In some embodiments, the emergency monitoring and management platform 130 may be configured in a processor and / or server. The processor and / or server may process data and / or information acquired from other platforms. Based on this data, information, and / or processing results, the processor and / or server may execute program instructions to perform one or more functions described in this application.

[0015] In some embodiments, the emergency monitoring and management platform 130 includes interconnected sub-platforms and a data center. The sub-platforms can process data and / or information acquired from the data center.

[0016] In some embodiments, the sub-platform includes at least one of an emergency prevention sub-platform, an emergency monitoring sub-platform, a risk prevention sub-platform, and an emergency response sub-platform.

[0017] The emergency prevention sub-platform refers to a management platform for assessing and preventing traffic emergency events.

[0018] The emergency monitoring sub-platform refers to a platform used for monitoring, collecting, and analyzing traffic emergency event data.

[0019] The risk prevention sub-platform refers to a platform for identifying potential risks, assessing risk levels, and implementing risk reduction strategies.

[0020] The emergency response sub-platform refers to a platform used for coordinating, dispatching, and executing emergency plans after a traffic emergency occurs.

[0021] In some embodiments, a data center includes a database, a data processing model library, and computing units.

[0022] Databases are used to collect, store, and manage large amounts of data related to emergency management. Examples include MySQL, PostgreSQL, InfluxDB, and Prometheus.

[0023] A data processing model library refers to a collection of data processing models used for emergency management data processing. In some embodiments, data processing models may include anomaly detection models, congestion prediction models, etc. For an explanation of anomaly detection models, see [link to documentation]. Figure 4 For details regarding the congestion prediction model, please refer to [link / reference]. Figure 5 And its contents.

[0024] A computing unit is a functional module used to perform arithmetic, logical, and other instruction operations. Computing units may include, but are not limited to, central processing units (CPUs).

[0025] In some embodiments, the system further includes an emergency monitoring user platform 110, an emergency monitoring service platform 120, an emergency monitoring sensor network platform 140, and an emergency monitoring object platform 150.

[0026] An emergency monitoring user platform refers to an interactive platform for emergency management personnel and the public. In some embodiments, the emergency monitoring user platform may include at least one user interaction device, such as a mobile phone or computer.

[0027] An emergency monitoring service platform refers to a platform that provides emergency monitoring services. In some embodiments, the emergency monitoring service platform can be configured as a server that can interact with emergency monitoring user platforms and emergency monitoring management platforms (such as data centers). For example, after discovering a traffic emergency, the public uses the emergency monitoring user platform to report the incident to the emergency monitoring service platform. The emergency monitoring service platform then reports the relevant incident and its impact assessment to the emergency monitoring management platform to facilitate response and decision-making by management departments.

[0028] An emergency monitoring sensor network platform refers to a platform for sensing and communicating emergency monitoring information in smart cities. It facilitates bidirectional data exchange and transmission between the emergency monitoring management platform (such as a data center) and the platform of the monitored entities. For example, an emergency monitoring sensor network platform may include communication equipment, servers, and various gateway devices. Smart city emergency monitoring information may include multi-dimensional data of the target tunnel, abnormal events, and upstream congestion conditions. For an explanation of multi-dimensional data of the target tunnel and upstream congestion conditions, please refer to [link to relevant documentation]. Figure 2 And its contents.

[0029] An emergency monitoring platform refers to an information processing platform for conducting safety supervision of various monitoring objects involved in emergency management. These monitoring objects can include non-motorized vehicle lanes, transportation hubs, public activity venues, etc. The platform can include various monitoring, sensing, and interactive devices, such as cameras, fire alarms, hazardous gas leak detectors, and environmental monitoring sensors.

[0030] In some embodiments of this specification, the smart city tunnel traffic emergency monitoring IoT big data model system 100 can form an information operation closed loop between various functional platforms and operate in a coordinated and regular manner under the unified management of the emergency monitoring management platform, thereby realizing the informatization and intelligentization of traffic emergency monitoring.

[0031] Figure 2 This is an exemplary flowchart of a smart city tunnel traffic emergency monitoring method according to some embodiments of this specification.

[0032] like Figure 2 As shown, process 200 includes the following steps. In some embodiments, process 200 may be executed by an emergency monitoring and management platform.

[0033] Step 210: Based on the multidimensional data of the target tunnel, construct the fusion feature vector of the vehicles in the target tunnel.

[0034] A target tunnel refers to a traffic tunnel monitored and managed by a smart city tunnel traffic emergency monitoring IoT big data model system. Examples include a main tunnel on an urban expressway or a highway tunnel.

[0035] Multidimensional data refers to a collection of data that comprehensively describes the traffic conditions of a road segment from multiple dimensions.

[0036] In some embodiments, multidimensional data may include timestamps, vehicle motion data, and vehicle identity information.

[0037] In some embodiments, vehicle motion data includes the vehicle position, instantaneous speed, acceleration, etc. of an individual vehicle.

[0038] A vehicle is a means of transportation that travels or stops on a certain road section.

[0039] Vehicle position can be represented based on the vehicle's three-dimensional coordinates.

[0040] In some embodiments, vehicle identification information includes vehicle type, license plate number, lane, etc.

[0041] Vehicle type refers to the category to which a vehicle belongs. For example, two-wheeled vehicles, passenger vehicles, and freight vehicles.

[0042] In some embodiments, multidimensional data can be acquired through various radars, sensors, and cameras.

[0043] In some embodiments, vehicle motion data can be acquired using millimeter-wave radar.

[0044] In some embodiments, vehicle identification information can be obtained through camera recognition.

[0045] For example, when a vehicle enters the camera's field of view, the radar captures the vehicle's position, instantaneous speed, and acceleration, while the camera identifies the vehicle's license plate number, vehicle type, and lane, and records the current timestamp.

[0046] A fused feature vector is a temporal feature vector formed by integrating multi-dimensional data. For example, the fused feature vector Fn for vehicle n is [timestamp, license plate number, vehicle type, vehicle position, instantaneous speed, acceleration, lane].

[0047] In some embodiments, the emergency monitoring and management platform can construct a fusion feature vector of vehicles within the target tunnel based on multidimensional data of the target tunnel.

[0048] In some embodiments, the step of the emergency monitoring and management platform constructing the fused feature vector may include: S1, the target tunnel is pre-installed with at least one radar and at least one camera operating on the same clock. At least one camera is set at a key location in the target tunnel. The emergency monitoring and management platform pre-records the monitoring space information of each camera. At least one radar continuously collects anonymous motion data of all moving targets and uploads it to the emergency monitoring and management platform in real time. At least one camera continuously identifies the vehicle identity information of all passing vehicles, records the corresponding timestamp, and uploads the vehicle identity information and timestamp to the emergency monitoring and management platform in real time.

[0049] The key locations are the pre-defined installation points for the cameras. For example, cameras can be installed at fixed intervals.

[0050] Monitoring spatial information refers to the data of the three-dimensional physical space that a camera can monitor.

[0051] Anonymous motion data refers to vehicle motion data that does not involve the vehicle's identity information. Anonymous motion data includes a timestamp, and at that timestamp, the vehicle's temporary ID, vehicle location, instantaneous speed, and acceleration.

[0052] A moving target refers to all moving entities within the target tunnel.

[0053] S2. After receiving the identity information and timestamp of a vehicle uploaded by the camera, the emergency monitoring and management platform queries the anonymous motion data based on the timestamp, and combines it with the pre-recorded monitoring space information to determine the only vehicle in the same monitoring space at the same time, and uniquely associates the anonymous motion data of the vehicle with the vehicle identity information.

[0054] S3, the emergency monitoring and management platform constructs the first fused feature vector of a vehicle based on the association between the vehicle's anonymous motion data and vehicle identity information; and updates the fused feature vector with all anonymous motion data related to the vehicle continuously uploaded by the radar, forming a fused feature vector sequence.

[0055] A fused feature vector sequence refers to an ordered set of fused feature vectors from multiple time series during the process of the same vehicle passing through a target tunnel.

[0056] Step 220: Identify abnormal events based on the fused feature vector and the congestion data of the target tunnel.

[0057] Congestion data is a dataset that characterizes the degree of traffic congestion on a particular road segment.

[0058] In some embodiments, congestion data may include lane flow, lane occupancy, vehicle queue length, average vehicle speed, and traffic density.

[0059] Lane flow refers to the number of vehicles passing through a certain road segment per unit of time.

[0060] Lane occupancy rate refers to the percentage of time a lane section on a road is occupied by vehicles within a preset time period.

[0061] In some embodiments, lane flow and lane occupancy can be obtained by sensors located on the lane cross-section.

[0062] Vehicle queue length refers to the length of the queue formed by vehicles on a certain road segment.

[0063] In some embodiments, the vehicle queue length can be obtained based on data collected by multiple sensors continuously positioned on the lane. For example, at least one sensor is placed at fixed intervals along a lane from start to finish. Each sensor has only two states: 0 and 1, where 0 indicates the lane is not occupied and 1 indicates the lane is occupied. The emergency monitoring and management platform continuously acquires the state of each sensor and, starting from the first sensor with a state of 1 closest to the start, identifies the number of consecutive sensors with a state of 1 along the direction of the finish line. The emergency monitoring and management platform multiplies the number of sensors by the fixed distance to determine the queue length.

[0064] Average vehicle speed refers to the average instantaneous speed of all vehicles within a certain road segment.

[0065] In some embodiments, the emergency monitoring and management platform can obtain the total number of vehicles and the instantaneous speed of each vehicle based on the fused feature vector of each vehicle, and determine the average vehicle speed as the ratio of the sum of the instantaneous speeds of the vehicles to the total number of vehicles.

[0066] Traffic density refers to the ratio of the number of vehicles to the area of ​​a given road segment.

[0067] In some embodiments, the emergency monitoring and management platform can obtain the vehicle location based on the fused feature vector of each vehicle, count the number of vehicles whose locations overlap with the location of the target tunnel, and determine the traffic density of the target tunnel as the ratio of the number of vehicles to the area.

[0068] An abnormal event refers to an event that occurs within a certain road segment and deviates from the normal traffic operation state.

[0069] In some embodiments, an abnormal event may include an abnormal type, an abnormal degree, an abnormal vehicle, and an abnormal location.

[0070] Anomaly type refers to the category to which an abnormal event belongs. For example, abnormal parking, illegal lane changing, speeding, and driving against traffic.

[0071] Anomaly severity refers to a quantitative indicator that characterizes the severity of an abnormal event. For example, anomaly severity is represented by a number from 0 to 100, with a higher value indicating a more severe anomaly.

[0072] Abnormal vehicles refer to the license plate numbers of vehicles involved in abnormal events.

[0073] An abnormal location point refers to the location information of an abnormal event that occurred in an abnormal vehicle, which can be represented based on three-dimensional coordinates and the lane in which it is located.

[0074] In some embodiments, the emergency monitoring and management platform can identify abnormal events in multiple ways based on the fused feature vectors and congestion data of the target tunnel.

[0075] In some embodiments, the emergency monitoring and management platform can construct a first target vector based on the fused feature vector of a vehicle in the target tunnel and congestion data; construct a first vector database based on a large amount of historical data from the data center, the first vector database including multiple first reference vectors and reference anomaly types and reference anomaly degrees corresponding to the first reference vectors; determine the first reference vector with the highest similarity to the first target vector and the corresponding highest similarity through vector matching; it is known that there are two cases: the highest similarity is greater than the similarity threshold and the highest similarity is less than or equal to the similarity threshold. If the highest similarity is greater than the similarity threshold, the reference anomaly type and reference anomaly degree corresponding to the first reference vector are determined as the anomaly type and anomaly degree of the first target vector. The determined anomaly type and anomaly degree, together with the license plate number and vehicle position in the fused feature vector of the vehicle, constitute an anomaly event; if the highest similarity is less than or equal to the similarity threshold, it is determined that the vehicle has no anomaly event.

[0076] In some embodiments, a first reference vector can be constructed by acquiring the fused feature vector of abnormal vehicles and congestion data when the abnormal event actually occurred in historical data; and a reference abnormality type and reference abnormality degree can be obtained by manually labeling the abnormality type and degree of the abnormal event.

[0077] Similarity thresholds can be preset based on historical experience.

[0078] In some embodiments, the emergency monitoring and management platform can also identify abnormal events through anomaly identification models. For more detailed information, please refer to [link / reference needed]. Figure 4 Related content.

[0079] Step 230: Based on abnormal events, congestion data, and traffic flow data around the target tunnel, predict the upstream congestion situation of the target tunnel.

[0080] Surrounding traffic flow data refers to traffic operation data of connecting road sections related to the target tunnel. For example, traffic operation data of the road section upstream of the target tunnel.

[0081] In some embodiments, surrounding traffic flow data may include average vehicle speed, traffic density, queue length at intersections, and regional traffic flow.

[0082] Average speed of a road segment refers to the average instantaneous speed of all vehicles on the connecting road segment.

[0083] Traffic density refers to the ratio of the number of vehicles to the area of ​​a road segment.

[0084] Intersection queue length refers to the length of vehicle queues at intersections located upstream of the target tunnel.

[0085] Regional flow refers to the number of vehicles passing through a connecting road segment per unit time. The method for obtaining regional flow is the same as that for lane flow in congestion data. The method for obtaining intersection queue length is the same as that for vehicle queue length in congestion data. The method for obtaining average vehicle speed on road segments is the same as that for average vehicle speed in congestion data. The method for obtaining traffic density on road segments is the same as that for traffic density in congestion data. For detailed explanation, please refer to the relevant content in step 220.

[0086] Upstream congestion refers to the predicted traffic congestion situation on the upstream section of the target tunnel within a future time period. This future time period can be pre-determined manually.

[0087] In some embodiments, upstream congestion includes predicted average vehicle speed and predicted queue length.

[0088] Predicted average vehicle speed refers to the average instantaneous speed of all vehicles within a predicted future time period.

[0089] Predicted queue length refers to the length of vehicle queues predicted for a future time period.

[0090] In some embodiments, the emergency monitoring and management platform can predict upstream congestion of the target tunnel in various ways based on abnormal events, congestion data, and traffic flow data around the target tunnel.

[0091] In some embodiments, the emergency monitoring and management platform can construct a second target vector based on abnormal events, congestion data, and surrounding traffic flow data; construct a second vector database based on a large amount of historical data from the data center, the second vector database including multiple second reference vectors and reference congestion conditions corresponding to the second reference vectors; determine the second reference vector with the highest similarity to the second target vector through vector matching, and determine the reference congestion condition corresponding to the second reference vector as the upstream congestion condition of the second target vector.

[0092] In some embodiments, a second reference vector can be constructed by acquiring the abnormal event, congestion data, and surrounding traffic flow data corresponding to the actual occurrence of the abnormal event from historical data; and the reference congestion situation corresponding to the second reference vector can be determined by acquiring the actual congestion situation of multiple upstream road sections of the target tunnel at the time of the abnormal event through cameras.

[0093] In some embodiments, the emergency monitoring and management platform can also predict upstream congestion based on a congestion prediction model. For more detailed information, please refer to [link / reference needed]. Figure 5 Related content.

[0094] Step 240: Generate signal control instructions, including travel time, based on upstream congestion conditions.

[0095] Signal control commands refer to a set of instructions that control the operating status and parameters of traffic lights. These commands are configured to control a target traffic light to maintain a target state for a specified duration; the target traffic light is a signal light on a critical road section heading towards a target tunnel.

[0096] Traffic lights refer to traffic indicators on lanes, such as red and green lights.

[0097] The target state refers to the traffic light being in a green state.

[0098] Critical road sections refer to upstream road sections where the degree of change in congestion data meets preset change conditions when an abnormal event occurs in the target tunnel. Examples include main roads directly connecting to the entrance of the target tunnel and road intersections within 3 kilometers of the entrance of the target tunnel.

[0099] The degree of change in congestion data refers to how congestion data changes when an abnormal event occurs in the target tunnel. For example, the degree to which average vehicle speed decreases.

[0100] Preset change conditions are the criteria for determining whether a certain road segment will affect the occurrence of abnormal events in the target tunnel. Preset change conditions can be determined manually. For example, preset change conditions could be an average vehicle speed decreasing to 50% of the minimum speed limit stipulated by the government transportation department or an average vehicle speed decreasing to 50% of the historical average vehicle speed.

[0101] In some embodiments, the signal control command is configured to control the traffic lights on at least one key road segment heading toward the target tunnel to display a green light for a specified duration.

[0102] In some embodiments, the signal control command may include the passage duration.

[0103] The duration of passage refers to the duration of the target state maintained in a traffic light, such as the display duration of a green light.

[0104] In some embodiments, the emergency monitoring and management platform can determine the passage time in various ways based on upstream congestion.

[0105] In some embodiments, the emergency monitoring and management platform can determine the passage time by querying a first preset table based on the upstream congestion situation.

[0106] The first preset table is a mapping table of upstream congestion and travel time.

[0107] The first preset table can be constructed using historical data or historical experience.

[0108] In some embodiments, the actual upstream congestion situation when historical abnormal events occurred in the target tunnel can be obtained; the adjusted travel time to meet preset requirements after reducing the green light duration of multiple key upstream road sections leading to the target tunnel can be obtained; the adjusted travel time can be mapped to the actual upstream congestion situation, and a first preset table can be constructed. The adjusted travel time refers to the travel time of vehicles that can improve the congestion in the target tunnel after adjusting the green light duration.

[0109] The preset requirement is the criterion for judging whether travel time can improve congestion. The actual upstream congestion situation is the upstream congestion situation corresponding to the time point when the abnormal event occurred in history.

[0110] In some embodiments, the preset requirements may include an average vehicle speed on the critical road segment being greater than a speed threshold and a queue length on the critical road segment being less than a queue length threshold. The speed threshold and queue length threshold are set based on human experience.

[0111] In some embodiments, the emergency monitoring and management platform can also determine the upstream congestion situation based on the upstream congestion situation and the distribution of congestion impact on the target tunnel. For further details, see [link to relevant documentation]. Figure 3 Related content.

[0112] Step 250: According to the signal control instruction, control the target traffic light to maintain the target state for the duration of passage.

[0113] In some embodiments, the emergency monitoring and management platform can control the target traffic lights to maintain the target state for the duration of passage according to signal control instructions.

[0114] For example, when the signal control instruction is [critical road segment A, passage time 60 seconds], the emergency monitoring and management platform controls the green light of critical road segment A to remain for 60 seconds.

[0115] Some embodiments of this specification can accurately identify minor abnormal events within a target tunnel and predict in advance the trend and scope of congestion spreading to upstream road sections; they can also control traffic lights on key roads to proactively restrict merging traffic before severe congestion occurs, thereby achieving a shift from "passively responding to congestion" to "proactively preventing congestion," shortening the reaction time from the occurrence of an event to the implementation of traffic management; and they can avoid delays and misjudgments caused by manual intervention, thereby reducing the risk of regional traffic paralysis caused by sudden events in the target tunnel.

[0116] In some embodiments, the smart city tunnel traffic emergency monitoring method 200 may further include determining a detour route based on upstream congestion and surrounding traffic flow data; generating a detour prompt instruction based on the detour route; controlling remote information boards on key road sections to display detour prompts based on the detour prompt instruction; and controlling vehicle-mounted terminals within key road sections to display detour prompts.

[0117] Detour routes refer to alternative routes planned for vehicles to avoid congestion, accidents, etc. For example, if a main road is congested, a secondary road can be used as a detour route.

[0118] In some embodiments, the emergency monitoring and management platform can determine detour routes based on upstream congestion conditions and surrounding traffic flow data. For example, the platform can identify a target tunnel whose upstream congestion meets preset congestion conditions as impending congestion; identify at least one connecting road segment upstream of the target tunnel whose surrounding traffic flow data meets preset smooth flow conditions as a detour route; and upload the detour route to a third-party map navigation service provider via a dedicated API interface. The third-party map navigation service provider's server calculates the detour time and returns the detour time and route to the emergency monitoring and management platform via the API interface.

[0119] Preset congestion and smooth traffic conditions can be determined based on historical experience. For example, a preset congestion condition is a predicted average vehicle speed of less than 15 km / h. A preset smooth traffic condition is an average vehicle speed of more than 40 km / h.

[0120] Detour instructions are a set of instructions conveyed to drivers, monitoring personnel, etc. For example, "Congestion ahead, please exit the main road at intersection 3."

[0121] In some embodiments, the emergency monitoring and management platform can generate detour prompts based on the detour route. For example, based on the detour time and route returned by the API interface, it can send a detour prompt to the driver: "Congestion ahead, please continue along the current road for 50 meters and then turn right."

[0122] Remote information boards are display devices installed on roads to show traffic information. For example, large display screens installed at intersections of main roads.

[0123] A vehicle-mounted terminal is a terminal device installed inside a vehicle that receives instructions from an emergency monitoring and management platform. Examples include dashcams and in-vehicle navigation systems.

[0124] In some embodiments, the emergency monitoring and management platform can control remote information boards on key road sections to display detour prompts and control vehicle-mounted terminals within key road sections to display detour prompts based on detour prompt instructions.

[0125] For example, the emergency monitoring and management platform will simultaneously send the detour prompt "Congestion ahead, please continue driving along the current road for 50 meters and then turn right" to the large display screens in the lane and the vehicle's in-vehicle navigation system.

[0126] Some embodiments in this specification, based on predicting congestion and adjusting traffic lights, further combine real-time traffic conditions to actively generate detour routes and release guidance information through multiple channels such as information boards and vehicle terminals. This achieves a leap from passive control of "congestion at the source" to proactive guidance of "bypassing congestion points," providing drivers with clear and feasible alternatives and significantly improving the overall traffic efficiency and emergency response capabilities of the regional road network.

[0127] Figure 3 This is an exemplary schematic diagram illustrating the determination of passage duration according to some embodiments of this specification.

[0128] In some embodiments, the emergency monitoring and management platform can also determine the travel time 360 ​​based on the upstream congestion situation 350 and the congestion impact distribution of the target tunnel 340.

[0129] The congestion impact distribution 340 refers to the degree of negative impact of congestion in the target tunnel on key upstream road sections.

[0130] In some embodiments, the congestion impact distribution includes key road segments and the degree to which those key road segments are affected by congestion at the target tunnel. For example, it is represented as key road segment A (main road, directly connected to the tunnel), with the degree of impact... = 0.9; Key road segment B (secondary arterial road, indirectly merging), impact level = 0.6.

[0131] In some embodiments, the emergency monitoring and management platform can determine the congestion impact distribution of the target tunnel in various ways. For example, the platform can use historical data to statistically analyze the actual speed change of each key road segment when an abnormal event occurs in the target tunnel; map the speed decrease to the corresponding impact level of the key road segment to obtain the congestion impact distribution. The more significant the speed decrease in a key road segment, the higher the impact level, and vice versa.

[0132] In some embodiments, the emergency monitoring and management platform can also determine the congestion impact distribution 340 of the target tunnel based on the upstream road distribution characteristics 310, tunnel characteristics 320, and traffic flow characteristics 330 of the target tunnel.

[0133] For detailed information on the distribution characteristics of upstream roads, tunnels, traffic flow, and the determination of the distribution of congestion impacts, please refer to [link to relevant documentation]. Figure 5 Related content.

[0134] Some embodiments in this specification, by combining the physical characteristics of the target tunnel with real-time traffic flow characteristics, can dynamically determine the potential impact range and distribution of congestion events.

[0135] In some embodiments, the emergency monitoring and management platform can determine the travel time based on upstream congestion and the distribution of congestion impact on the target tunnel.

[0136] In some embodiments, travel time can be calculated based on upstream congestion on key road segments and the distribution of congestion impact on the target tunnel.

[0137] For example, the passage time can be determined sequentially using formulas (1), (2), and (3): (1) In formula (1), The congestion score represents the overall congestion level of key road sections; The vehicle speed is scored, representing the degree of slowdown in traffic flow on key road sections; Queueing scores represent the severity of vehicle queuing at key road sections in future timeframes.

[0138] The ratio of the critical congestion speed to the predicted average speed on key road sections is calculated. When the predicted average speed is 0, or the aforementioned ratio is greater than 1, the default setting is... Revised to 1. The congestion threshold speed refers to the critical speed at which congestion is about to occur. The congestion threshold speed can be set based on human experience.

[0139] The maximum warning length is obtained by calculating the ratio of the future queue length to the maximum warning length of the critical road segment. The maximum warning length is the length of the road segment.

[0140] (2) In formula (2), The pressure index is used to characterize the pressure on key road sections to regulate travel time. The impact level of critical road sections represents the degree to which critical road sections are affected by abnormal events in the target tunnel.

[0141] (3) In formula (3), For travel duration; Characterizes the duration of a green light under normal circumstances; It represents the shortest green light duration to ensure safe passage. , Settings based on human experience. For example, It lasts for 60 seconds. It lasts for 15 seconds.

[0142] For the methods used to determine the predicted average vehicle speed and predicted queue length, please refer to [link / reference]. Figure 1 For related explanations, please refer to the following: Regarding the method for determining the degree of impact, see... Figure 5 Related explanations.

[0143] Some embodiments in this specification determine the duration of traffic light passage by analyzing the distribution of congestion impact. This enables a shift from "extensive" regional traffic control to "precise" differentiated regulation. Consequently, it can accurately identify the upstream key road sections that contribute the most to tunnel congestion and implement targeted and forceful traffic reductions on them. At the same time, it imposes only minor restrictions on secondary road sections with less impact. This not only greatly improves the efficiency and targeting of source traffic control but also avoids unnecessary interference to surrounding traffic caused by "one-size-fits-all" regulation, achieving a balance between ensuring the effectiveness of emergency traffic management and maintaining regional traffic order.

[0144] Figure 4 This is an exemplary schematic diagram of an anomaly recognition model according to some embodiments of this specification.

[0145] In some embodiments, the emergency monitoring and management platform can also determine abnormal events 440 based on the fused feature vector 410 and congestion data 420 through the anomaly identification model 430.

[0146] Anomaly identification model 430 is a model used to identify anomalous events.

[0147] The anomaly detection model is a machine learning model. For example, any one or a combination of Long Short-Term Memory (LSTM) networks or other custom model structures.

[0148] The anomaly detection model takes into account the fused feature vector of a single vehicle and congestion data, and outputs the anomaly type and degree corresponding to that vehicle.

[0149] For a detailed explanation of fusing feature vectors, congestion data, anomaly types, and anomaly severity, please refer to [link to relevant documentation]. Figure 1 Related content.

[0150] In some embodiments, the anomaly detection model can be trained using multiple first training samples with a first label. For example, multiple first training samples with a first label can be input into an initial detection model. A loss function is constructed using the first labels and the results of the initial detection model. The parameters of the initial detection model are iteratively updated based on the loss function using gradient descent or other methods. The model training is complete when preset conditions are met, resulting in a trained anomaly detection model. The preset conditions may be loss function convergence, the number of iterations reaching a threshold, etc.

[0151] The first training sample may include historical fused feature vectors and historical congestion data.

[0152] The first label can include the actual anomaly type and the actual anomaly degree corresponding to the actual occurrence of the anomaly event under the first training sample.

[0153] In some embodiments, the first training sample and the first label may be obtained based on historical data.

[0154] For example, when acquiring historical data on actual abnormal events, the fused feature vector of a single vehicle and congestion data are used to manually label the actual abnormal type and degree of the abnormal event, thus obtaining the first label.

[0155] In some embodiments of this specification, the anomaly recognition model can automatically learn and understand the behavior patterns of a vehicle over a continuous period of time, and more sensitively capture subtle and critical changes in vehicle motion data, thereby accurately identifying complex anomaly events that are difficult to judge based on instantaneous states alone. This improves the accuracy and comprehensiveness of anomaly event recognition, making emergency response more timely and effective.

[0156] In some embodiments, the emergency monitoring and management platform can also generate lane control instructions based on the lane corresponding to the abnormal event; based on the lane control instructions, control the lane indicator located upstream of the lane to display a no-passing signal, and activate the traffic lights at the ramp entrance of the lane to display in a preset display state.

[0157] For information on how to determine the lane corresponding to an abnormal event, please refer to [link / reference]. Figure 1 The relevant content of step 220.

[0158] Lane control instructions refer to a set of instructions that control the passage of traffic in a lane. For example, lane X is closed to traffic, while lane Y is open to traffic.

[0159] In some embodiments, lane control instructions may include instructions to control lane indicators and instructions to control ramp traffic lights.

[0160] Lane indicators are visual devices that indicate whether a lane is passable. For example, a display screen.

[0161] A prohibition signal is a specific signal displayed by a lane indicator that tells vehicles that passage in that lane is prohibited. Examples include a red "×" or other arbitrary graphic symbol.

[0162] The preset display state is manually preset, which is the display state of the lane indicator.

[0163] In some embodiments, the preset display status may include allow passage and prohibit passage.

[0164] In some embodiments, the emergency monitoring and management platform can generate lane control instructions based on the lane corresponding to the abnormal event; based on the lane control instructions, it can control the lane indicator located upstream of the lane to display a no-passing signal, and start the traffic lights at the ramp entrance of the lane to display in a preset display state.

[0165] For example, if an abnormal event occurs in lane I, the emergency monitoring and management platform generates a lane control instruction to prohibit passage in lane I, and the lane indicator upstream of that lane displays a red "×". If the target tunnel has a ramp, the emergency monitoring and management platform generates a lane control instruction to "release one vehicle every 15 seconds", and the traffic lights at the ramp entrance alternate between a green light for 3 seconds and a red light for 15 seconds.

[0166] Some embodiments in this specification refine the granularity of emergency response from the macroscopic road segment level to the lane level. When a specific lane becomes congested, the system can immediately and accurately close the upstream entrance of that lane and guide the traffic flow, effectively preventing secondary accidents. In addition, when dealing with mainline congestion, the system actively controls the merging flow by activating ramp traffic lights, realizing dynamic regulation of the tunnel traffic "microcirculation" and effectively delaying or eliminating the accumulation of mainline congestion.

[0167] Figure 5 This is an exemplary schematic diagram illustrating the determination of upstream congestion according to some embodiments of this specification.

[0168] In some embodiments, the emergency monitoring and management platform is further configured to: construct a congestion map 520 based on congestion data 420, abnormal events 440, and target traffic flow data 510; and predict upstream congestion conditions 350 based on the congestion map 520 using a congestion prediction model 530. For explanations regarding congestion data, abnormal events, and upstream congestion conditions, please refer to... Figure 2 And its contents.

[0169] Target traffic flow data refers to the traffic flow data surrounding the target tunnel within a preset range.

[0170] The preset range refers to the preset maximum surrounding area that is farthest from the target tunnel. In some embodiments, the preset range may be determined manually. In some embodiments, the preset range may also be determined based on the upstream road distribution characteristics, tunnel characteristics, and traffic flow characteristics of the target tunnel.

[0171] In some embodiments, the emergency monitoring and management platform is further configured to: determine the congestion impact distribution of the target tunnel based on the upstream road distribution characteristics, tunnel characteristics, and traffic flow characteristics; and determine a preset range based on the congestion impact distribution.

[0172] The upstream road distribution characteristics refer to the distribution characteristics of the roads upstream of the target tunnel. Upstream roads are those whose travel direction merges into the target tunnel.

[0173] In some embodiments, the upstream road distribution characteristics include the number of roads merging into the target tunnel and the number of lanes. Different upstream roads may have different numbers of lanes. For example, upstream roads may include roads merging into the target tunnel after going straight and roads merging into the target tunnel after turning right, wherein the roads merging into the target tunnel after going straight have 2 lanes and the roads merging into the target tunnel after turning right have 1 lane. The emergency monitoring and management platform can determine the upstream road distribution characteristics based on pre-acquired survey reports.

[0174] Tunnel characteristics refer to the relevant features of a target tunnel. Examples include tunnel width, the number of lanes within the tunnel, and tunnel length. In some embodiments, the emergency monitoring and management platform can determine tunnel characteristics based on pre-acquired survey reports.

[0175] Traffic flow characteristics refer to features related to traffic flow within the target tunnel. Examples include traffic density and traffic type.

[0176] Traffic density refers to the density of vehicles within the target tunnel. In some embodiments, traffic density can be obtained based on congestion data.

[0177] Traffic flow type refers to the types of vehicles and the number of different vehicle types within the target tunnel. For example, cars (15 vehicles), trucks (13 vehicles), etc. In some embodiments, the emergency monitoring and management platform can obtain vehicle types based on multidimensional data of the target tunnel and count the number of vehicles corresponding to each vehicle type.

[0178] For an explanation of the distribution of congestion impacts, please refer to [link / reference]. Figure 3 And its contents.

[0179] In some embodiments, the emergency monitoring and management platform can determine the congestion impact distribution of a target tunnel based on the upstream road distribution characteristics, tunnel characteristics, and traffic flow characteristics. For example, the platform can construct a matching vector based on these characteristics, then query a third reference vector in a third vector database that matches the matching vector. The label of this third reference vector is then used to determine the congestion impact distribution of the target tunnel. Matching conditions include the highest similarity between vectors. Vector similarity is negatively correlated with vector distance. Vector distance includes Euclidean distance, etc.

[0180] In some embodiments, the third vector database can be pre-configured based on historical data, containing multiple third reference vectors and labels corresponding to each third reference vector. For example, the emergency monitoring and management platform can acquire historical data from the occurrence of historical abnormal events, and construct corresponding third reference vectors based on the historical upstream road distribution characteristics, historical tunnel characteristics, and historical traffic flow characteristics of each historical target tunnel at the time of each historical abnormal event. The historical congestion impact distribution corresponding to each set of historical upstream road distribution characteristics, historical tunnel characteristics, and historical traffic flow characteristics is used as the label of the corresponding third reference vector. For example, the label of the third reference vector can be {(critical road segment A, A1), (critical road segment B, B1)}, where A1 and B1 represent the degree of impact corresponding to critical road segments A and B of the historical target tunnel, respectively.

[0181] In some embodiments, the label of the third reference vector may be determined manually, including: obtaining congestion data of the upstream road of the historical target tunnel corresponding to the third reference vector, determining the degree of change when the change trend is a decrease in average speed within a preset historical period, identifying road segments whose degree of change meets preset change conditions as key road segments, and determining the degree of change corresponding to the key road segment as the degree of influence.

[0182] For example, if the average speed on a certain road segment decreases by 60%, meaning the congestion data changes by 60%, exceeding the preset change condition (e.g., an average speed decrease of 50%), this road segment is designated as a critical segment, with a corresponding impact level of 0.6. Conversely, if the average speed on a certain road segment increases, it indicates smooth traffic flow on that segment, meaning it is not a critical segment. For an explanation of the degree of congestion data change, critical segments, and preset change conditions, please refer to [link to relevant documentation]. Figure 2 And its contents.

[0183] In some embodiments, the emergency monitoring and management platform can determine the distance between the critical road segment furthest from the target tunnel in the congestion impact distribution and the target tunnel as the preset range of the target tunnel.

[0184] In some embodiments of this specification, by combining tunnel characteristics with real-time traffic flow characteristics, the potential impact range and distribution of congestion events can be dynamically determined. This allows for the delineation of a reasonably sized preset range for emergency monitoring, enabling precise focus on key road sections truly affected by abnormal events. This improves the targeting and effectiveness of emergency response and avoids resource waste due to an excessively large monitoring range or control oversights due to an excessively small monitoring range.

[0185] A congestion graph (520) refers to a graph characterizing traffic congestion conditions and can be used to determine upstream congestion. In some embodiments, the congestion graph consists of nodes and edges. For example... Figure 5 As shown, the congestion graph 520 includes multiple nodes (e.g., A, B, A1, B1, A2) and multiple directed edges, including unidirectional edges and bidirectional edges.

[0186] In some embodiments, the nodes in a congestion map region may include a central node and diffusion nodes.

[0187] In some embodiments, the emergency monitoring and management platform can determine the central node, the diffusion node, and the edges connecting the nodes based on congestion data 420, abnormal events 440, and target traffic flow data 510, in order to construct a congestion map.

[0188] A central node is a node that can be represented as a target tunnel. In some embodiments, the target tunnel can be divided into multiple tunnel critical segments, and each tunnel critical segment corresponds to a central node.

[0189] A critical tunnel segment refers to a key section within the target tunnel. For example, a target tunnel can be divided into multiple critical tunnel segments such as the "entrance segment," "middle segment," "exit segment," and "ramp merging segment." Each critical tunnel segment can correspond to a central node, for example... Figure 5 A, A1, A2 in the example.

[0190] In some embodiments, the node attributes of the central node may include data related to critical tunnel segments. For example, the corresponding tunnel characteristics, congestion data, and event severity index for the critical tunnel segments. See [link to documentation] for an explanation of congestion data. Figure 2 For details regarding its contents and descriptions of tunnel features, please refer to [link / reference]. Figure 5 The content above.

[0191] An event severity index is a numerical value used to characterize the severity of an abnormal event. In some embodiments, the event severity index can be a normalized value between 0 and 1. The emergency monitoring and management platform can determine the event severity index based on the real-time status of the abnormal event and according to preset rules. For example, preset rules may include: no abnormal event, event severity index is 0; abnormal event exists in a single lane, event severity index is 0.5; abnormal event exists in multiple lanes, event severity index is 0.8; tunnel is completely blocked, event severity index is 1.0. Here, a complete tunnel blockage can refer to the average speed on the critical road section being less than an average speed threshold, which can be determined based on empirical presets. An abnormal event exists in multiple lanes, meaning that abnormal events occur simultaneously in two or more lanes.

[0192] A diffusion node is a node that can be represented as a critical segment of the upstream road of the target tunnel. In some embodiments, the upstream road of the target tunnel may also be divided into multiple critical segments, with each critical segment of the upstream road corresponding to a diffusion node, for example... Figure 5 The nodes B, B1, and B2 in the diagram represent the diffusion nodes. Node attributes can include data related to upstream road segments. For example, traffic flow data corresponding to key segments of upstream roads. See [link to documentation] for an explanation of traffic flow data. Figure 2 And its contents.

[0193] In some embodiments, multiple nodes may be connected by edges. Edges may include connecting edges and logical edges.

[0194] Connecting edges are directed edges that represent the direction of traffic flow between critical road segments. For example, such as... Figure 5 The solid lines shown are for reference only. In some embodiments, when there is a connection between two key road segments corresponding to two nodes (e.g., they are interconnected), the two nodes are connected based on the connecting edge. The direction of the edge represents the traffic flow direction. The traffic flow direction refers to the planned driving direction of the road. For example, as shown... Figure 5 As shown, A is the exit section, A1 is the middle section, and A2 is the entrance section. The traffic flow direction is A2-A1-A.

[0195] In some embodiments, connecting edges may exist between diffusion nodes, between diffusion nodes and the central node, and between central nodes. For example, such as Figure 5As shown, there is a connecting edge between diffusion node B and diffusion node B2, a connecting edge between center node A and diffusion node B, and a connecting edge between center node A and center node A1. The center node, representing the entrance or exit of the target tunnel, can be connected to the diffusion nodes.

[0196] In some embodiments, the characteristics of the connecting edge include variations in traffic capacity and number of lanes.

[0197] Traffic capacity refers to the maximum number of vehicles that can pass through a node's connection point per hour. In some embodiments, the connection point of a node may include intersections, ramps, etc. Traffic capacity can be obtained based on historical records.

[0198] Lane number change refers to the difference in the number of lanes between two nodes. For example, ... Figure 5 As shown, node A2 has 3 lanes, node A1 has 2 lanes, and the traffic flow direction is A2-A1, representing a 2 / 3 change in the number of lanes. Due to the narrowing of the lanes, vehicle density may increase, and the probability of abnormal events may also rise. In some embodiments, the change in the number of lanes can be determined based on tunnel characteristics. For an explanation of tunnel characteristics, please refer to [link to relevant documentation]. Figure 5 The content above.

[0199] A logical edge is a directed edge that represents the relationship between multiple nodes regarding congestion levels. For example, ... Figure 5 The dashed lines are shown. In some embodiments, when there is a causal relationship of congestion between two critical road segments corresponding to two nodes, the two nodes are connected based on a logical edge. The causal relationship of congestion refers to an abnormal event occurring at a downstream node, causing a change in the traffic situation at an upstream node. The direction of the logical edge is opposite to the direction of traffic flow, that is, from the downstream node to the upstream node. For example, as shown... Figure 5 As shown, if the traffic flow direction is A2-A1-A, then the direction of the logical edge is A-A1-A2.

[0200] In some embodiments, logical edges may exist between diffusion nodes and central nodes, and between central nodes themselves. For example, such as... Figure 5 As shown, there is a logical edge between central node A and diffusion node B, and a logical edge between central node A and central node A1. Logical edges cannot exist between diffusion nodes. Only the central node representing the entrance or exit of the target tunnel can have logical edges with diffusion nodes.

[0201] In some embodiments, logical edges are characterized by influence weights. Influence weights refer to the degree of influence a downstream node has on the traffic situation of an upstream node. Influence weights can be determined based on the degree of influence. For an explanation of influence degree, please refer to [link to relevant documentation]. Figure 3 and Figure 5 The content above.

[0202] In some embodiments, if there are both connecting edges and logical edges between two nodes, a bidirectional edge is formed, representing that there is both a connecting relationship and a causal relationship of congestion between the key road segments corresponding to the two nodes.

[0203] In some embodiments, the emergency monitoring and management platform is further configured to: determine the central node and the diffusion node based on the tunnel characteristics, congestion data and target traffic flow data of the target tunnel; and construct a congestion map based on the central node and the diffusion node.

[0204] In some embodiments, the emergency monitoring and management platform can determine the central node based on tunnel characteristics and congestion data. For example, the emergency monitoring and management platform can determine the node attributes of the key tunnel segment (i.e., the central node) based on the corresponding tunnel characteristics and congestion data of the key tunnel segment.

[0205] In some embodiments, the emergency monitoring and management platform can determine diffusion nodes based on target traffic flow data. For example, the emergency monitoring and management platform can determine the surrounding traffic flow data corresponding to key road segments of multiple upstream roads based on target traffic flow data, and each piece of surrounding traffic flow data corresponds to the node attributes of a key road segment (i.e., diffusion node) of an upstream road.

[0206] In some embodiments, the emergency monitoring and management platform constructs a congestion graph based on the central node, the diffusion nodes, and the first and second types of edges connecting the nodes.

[0207] The first type of edge refers to the logical edge between central nodes. For example, ... Figure 5 As shown, the logical edge A1A2 is between the central node A2 and the central node A1.

[0208] The second type of edge refers to the logical edge between the central node and the spreading nodes. For example, such as... Figure 5 As shown, the logical edge A2B is between the central node A2 and the diffusion node B.

[0209] In some embodiments, in the congestion graph, the edge features of the first type of edge and the second type of edge include influence weights, wherein the influence weight corresponding to the first type of edge is not less than the influence weight corresponding to the second type of edge.

[0210] like Figure 3 As described in the relevant content, the degree of influence of a downstream node on the traffic situation of an upstream node refers to the severity of the impact of an abnormal event of a downstream node on an upstream node. In some embodiments, the emergency monitoring and management platform can determine the degree of influence of the diffusion nodes (i.e., nodes in the upstream direction) connected by the second type of edge as the influence weight of that second type of edge. For details on the determination and explanation of the degree of influence, please refer to [link to relevant documentation]. Figure 3 The corresponding content.

[0211] like Figure 5 As shown, the central node (e.g., A1) located in the middle of the target tunnel cannot be connected to the diffusion node. That is, only one of the two central nodes connected by the first type of edge is connected to the diffusion node. Therefore, for a first type of edge (e.g., A1A2), the emergency monitoring and management platform can also determine the central node (e.g., A2) connected to the diffusion node within that first type of edge, obtain all second type edges (e.g., A2B) connected to that central node, add the largest influence weight among one or more second type edges to the regular item C, and use the sum as the influence weight of the first type of edge (e.g., A1A2). The regular item C can be determined by manual experience.

[0212] In some embodiments, the maximum value of the influence weight is 1. When the calculated influence weight is greater than 1, the emergency supervision and management platform determines the calculation result of the influence weight to be 1.

[0213] In some embodiments of this specification, by assigning influence weights to the edges between nodes in the congestion graph, the degree of influence of downstream road segments on upstream road segments can be intuitively displayed, which facilitates the subsequent determination of upstream congestion based on the congestion prediction model.

[0214] In some embodiments, the emergency monitoring and management platform can predict upstream congestion 350 based on the congestion map 520 and the congestion prediction model 530.

[0215] A congestion prediction model is a model used to predict upstream congestion. In some embodiments, the congestion prediction model can be a machine learning model. For example, a congestion prediction model can be a Graphical Neural Network (GNN), etc. In some embodiments, the input of the congestion prediction model can include a congestion map, and the output can include the upstream congestion situation of the target tunnel.

[0216] In some embodiments, the emergency monitoring and management platform can train and construct a congestion prediction model using a large number of training samples and corresponding training labels. In some embodiments, the platform can acquire a training dataset, which includes multiple second training samples and corresponding second training labels for each sample. The platform can then perform multiple iterations to obtain the trained congestion prediction model. The training of the congestion prediction model is similar to that of the anomaly detection model; for details, please refer to the training instructions for the anomaly detection model.

[0217] In some embodiments, the emergency monitoring and management platform can identify historical anomalies and their corresponding congestion data at a first point in time, as well as target traffic flow data within a preset range, from historical data to construct a historical congestion map, which is then used as a second training sample. The platform can also obtain the actual congestion situation of all key road sections upstream of the target tunnel corresponding to each second training sample at the second point in time, and use the actual congestion situation as the second training label based on manual annotation. Both the first and second points in time are historical points in time, with the first point preceding the second point in time. The interval between the two points can be preset based on manual experience, for example, 5 minutes.

[0218] In some embodiments of this specification, by using trained machine learning models, the traffic network can be treated as a whole, interconnected system, rather than simply analyzing isolated road segment data. The congestion prediction model, by learning the propagation patterns of upstream congestion within the road network topology, can accurately predict how and at what speed abnormal events at a target tunnel affect the various upstream road segments connected to the target tunnel. This makes the prediction results closer to the actual congestion spread process, thereby enabling the development of more forward-looking and holistic mitigation strategies for abnormal events.

[0219] The basic concepts have been described above. Obviously, for those skilled in the art, the detailed disclosure above is merely illustrative and does not constitute a limitation of this specification. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements, and corrections to this specification. Such modifications, improvements, and corrections are suggested in this specification and therefore remain within the spirit and scope of the exemplary embodiments described herein.

[0220] Finally, it should be understood that the embodiments described in this specification are merely illustrative of the principles of the embodiments described herein. Other variations may also fall within the scope of this specification. Therefore, alternative configurations of the embodiments described herein are intended to be illustrative rather than limiting, and should be considered consistent with the teachings of this specification. Accordingly, the embodiments described herein are not limited to those explicitly introduced and described herein.

Claims

1. A smart city tunnel traffic emergency monitoring IoT large-scale model system, characterized in that, This includes an emergency monitoring and management platform, which is configured as follows: Based on the multidimensional data of the target tunnel, a fusion feature vector of the vehicles in the target tunnel is constructed; Based on the fused feature vector and the congestion data of the target tunnel, abnormal events are identified through an anomaly identification model, which is a machine learning model. The abnormal events include anomaly type, anomaly degree, abnormal vehicles, and abnormal location points. The anomaly type includes abnormal parking, illegal lane changing, speeding, and driving in the wrong direction. Based on the abnormal events, the congestion data, and the traffic flow data around the target tunnel, predict the upstream congestion situation of the target tunnel; Based on the upstream congestion situation, a signal control instruction including a passage duration is generated. The signal control instruction is configured to control a target traffic light to maintain a target state for the passage duration. The target traffic light is a traffic light on a key road segment heading towards the target tunnel. The passage duration is determined based on the upstream congestion situation and the congestion impact distribution of the target tunnel. The emergency monitoring and management platform is further configured as follows: Based on the congestion data, the abnormal events, and the target traffic flow data, central nodes, diffusion nodes, and the edges connecting the nodes are determined to construct a congestion map. Logical edges exist between the diffusion nodes and the central nodes, and between the central nodes themselves. No logical edges exist between the diffusion nodes. The direction of the logical edges is opposite to the traffic flow direction. The logical edges between the central nodes are classified as first-type edges, and the logical edges between the central nodes and the diffusion nodes are classified as second-type edges. The edge characteristics of the first-type and second-type edges include influence weights, and the influence weight corresponding to the first-type edge is not less than the influence weight corresponding to the second-type edge. The target traffic flow data is the surrounding traffic flow data within a preset range of the target tunnel. Based on the congestion map, the upstream congestion situation is predicted using a congestion prediction model, which is a machine learning model. The emergency monitoring and management platform is further configured as follows: Based on the upstream road distribution characteristics, tunnel characteristics, and traffic flow characteristics of the target tunnel, the congestion impact distribution of the target tunnel is determined; The preset range is determined based on the distribution of congestion impact; The emergency monitoring and management platform is configured as follows: Based on the upstream congestion situation and the surrounding traffic flow data, a detour route is determined; Based on the detour route, generate a detour prompt instruction; According to the detour prompt instruction, control the remote information board on the key road section to display the detour prompt, and control the vehicle terminal in the key road section to display the detour prompt; The emergency monitoring and management platform is configured as follows: Based on the lane corresponding to the abnormal event, a lane control command is generated; According to the lane control command, the lane indicator located upstream of the lane is controlled to display a no-entry signal, and the traffic lights at the ramp entrance of the lane are activated to display in a preset state.

2. A smart city tunnel traffic emergency monitoring method, characterized in that, Based on the emergency monitoring and management platform, the method includes: Based on the multidimensional data of the target tunnel, a fusion feature vector of the vehicles in the target tunnel is constructed; Based on the fused feature vector and the congestion data of the target tunnel, abnormal events are identified through an anomaly identification model, which is a machine learning model. The abnormal events include anomaly type, anomaly degree, abnormal vehicles, and abnormal location points. The anomaly type includes abnormal parking, illegal lane changing, speeding, and driving in the wrong direction. Based on the abnormal events, the congestion data, and the traffic flow data around the target tunnel, predict the upstream congestion situation of the target tunnel; Based on the upstream congestion situation, generate signal control instructions including travel duration; According to the signal control command, the target traffic light is controlled to maintain the target state for the specified passage duration. The target traffic light is a traffic light on a key road section heading towards the target tunnel. The passage duration is determined based on the upstream congestion situation and the congestion impact distribution of the target tunnel. The step of predicting the upstream congestion situation of the target tunnel based on the abnormal event, the congestion data, and the traffic flow data surrounding the target tunnel includes: Based on the congestion data, the abnormal events, and the target traffic flow data, central nodes, diffusion nodes, and the edges connecting the nodes are determined to construct a congestion map. Logical edges exist between the diffusion nodes and the central nodes, and between the central nodes themselves. No logical edges exist between the diffusion nodes. The direction of the logical edges is opposite to the traffic flow direction. The logical edges between the central nodes are classified as first-type edges, and the logical edges between the central nodes and the diffusion nodes are classified as second-type edges. The edge characteristics of the first-type and second-type edges include influence weights, and the influence weight corresponding to the first-type edge is not less than the influence weight corresponding to the second-type edge. The target traffic flow data is the surrounding traffic flow data within a preset range of the target tunnel. Based on the congestion map, the upstream congestion situation is predicted using a congestion prediction model, which is a machine learning model. The method further includes: Based on the upstream road distribution characteristics, tunnel characteristics, and traffic flow characteristics of the target tunnel, the congestion impact distribution of the target tunnel is determined; The preset range is determined based on the distribution of congestion impact; The method further includes: Based on the upstream congestion situation and the surrounding traffic flow data, a detour route is determined; Based on the detour route, generate a detour prompt instruction; According to the detour prompt instruction, control the remote information board on the key road section to display the detour prompt, and control the vehicle terminal in the key road section to display the detour prompt; The method further includes: Based on the lane corresponding to the abnormal event, a lane control command is generated; According to the lane control command, the lane indicator located upstream of the lane is controlled to display a no-entry signal, and the traffic lights at the ramp entrance of the lane are activated to display in a preset state.

Images