Meteorological disaster early warning system based on C-V2X and edge calculation

Abstract

The invention relates to a meteorological disaster early warning system based on C-V2X and edge calculation, and belongs to the technical field of traffic meteorological monitoring. The system comprises an edge calculation module and a C-V2X communication module. The edge calculation module collects video image information and vehicle state information and carries out multi-mode perception fusion, meteorological disaster identification is carried out on fused data through an embedded artificial intelligence model, and then direct communication between vehicles is realized based on a C-V2X technology. Early warning information is accurately delivered to subsequent vehicles which are on the same route and run in the same direction through direct communication between the vehicles, a driver is reminded to drive carefully or take necessary measures in advance, and the road traffic safety is improved. The vehicle-mounted edge calculation and artificial intelligence technology is utilized, real-time monitoring and early warning of the road weather condition are achieved, the safety and passing efficiency of road traffic are improved, and the method has wide application prospects.

Description

Technical Field

[0001] This invention belongs to the field of traffic meteorological monitoring technology, specifically involving a meteorological disaster early warning system based on C-V2X and edge computing. Background Technology

[0002] With the development of intelligent transportation systems, the monitoring and dissemination of real-time traffic and meteorological disaster information has become particularly important. Traditional meteorological monitoring mainly relies on fixed roadside meteorological observation equipment (hereinafter referred to as "roadside equipment"). These roadside devices are limited in number and sparsely distributed, making it difficult to reflect the actual weather conditions on the road in a timely and comprehensive manner. In addition, roadside equipment is exposed to the elements, making it susceptible to environmental influences and resulting in high maintenance costs.

[0003] Currently, traffic meteorological disaster monitoring systems based on roadside equipment mainly suffer from the following problems: First, insufficient monitoring accuracy. The spacing between roadside equipment is typically 10-30 kilometers, which cannot meet the needs of refined meteorological monitoring, especially in complex terrain areas where meteorological conditions can vary significantly within a few hundred meters. Second, poor real-time performance. Meteorological data collected by roadside equipment needs to be processed by the traffic management center before being released, resulting in a 3-5 minute information delay, which cannot meet the real-time early warning needs for driving safety. Third, high cost and difficult maintenance. The construction cost and annual maintenance cost of each roadside device are high, and they are easily damaged by severe weather. Fourth, limited coverage. Roadside equipment can only monitor meteorological conditions within a limited area around the station and cannot obtain actual road traffic conditions.

[0004] Besides roadside equipment-based solutions, existing vehicle-mounted meteorological monitoring technologies also have limitations.

[0005] For example, CN 113727304A discloses an emergency vehicle early warning system and its early warning method based on a 5G communication architecture. The system architecture relies on roadside facilities, and its perception fusion processing is completed in a 5G mobile edge computing unit on the roadside. The roadside unit then sends the processing results to the vehicle.

[0006] CN 106781697A discloses a vehicle-mounted real-time adverse weather perception and collision avoidance warning method. Its core lies in using perceived weather information as input parameters for a collision avoidance warning module to improve the collision avoidance performance of a single vehicle in dangerous situations and ultimately achieve control over the vehicle. While this technology achieves vehicle-mounted weather perception, its core purpose is to assist the vehicle's collision avoidance system, using weather information as a parameter to optimize vehicle control, rather than building a wide-area collaborative warning network.

[0007] CN 119169769A discloses a road traffic natural disaster early warning and prevention system based on AIoT. This system architecture is built upon fixed road infrastructure, and its edge layer module consists of various fixed sensors deployed along the road and intelligent natural disaster monitoring terminals. Its drawbacks are that the system heavily relies on expensive fixed infrastructure investment, resulting in high construction and maintenance costs. Furthermore, its monitoring range is limited by the sensor deployment locations, inevitably leading to numerous monitoring blind spots in the vast road network. This makes it impossible to achieve seamless coverage of road weather conditions and effective early warning of sudden, small-scale disasters.

[0008] In addition, existing technologies of this kind either rely on earlier communication technologies such as DSRC for information exchange, which have limited transmission efficiency and reliability; or they use a general area broadcasting method, which cannot achieve accurate delivery of warning information and is prone to causing information interference to irrelevant vehicles, thus greatly reducing the effectiveness of the warning.

[0009] Therefore, overcoming the shortcomings of the existing technologies is an urgent problem to be solved in the field of traffic meteorological monitoring technology. Summary of the Invention

[0010] The purpose of this invention is to address the shortcomings of existing technologies and provide a vehicle-mounted edge computing system for real-time monitoring and early warning of traffic and meteorological disasters based on C-V2X technology.

[0011] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0012] A meteorological disaster early warning system based on C-V2X and edge computing includes vehicle-mounted equipment installed on a vehicle; the vehicle-mounted equipment includes an edge computing module and a C-V2X communication module.

[0013] The edge computing module is used to collect video image information and vehicle status information of the vehicle, perform multimodal perception fusion on the collected video image information and vehicle status information, and process the fused data through the embedded artificial intelligence model built into the edge computing module to identify the type and level of meteorological disasters in the vehicle's driving environment and generate corresponding early warning information.

[0014] The C-V2X communication module, connected to the edge computing module, is used to send warning information to following vehicles traveling in the same direction on the same route through direct inter-vehicle communication. After receiving the warning information through its C-V2X communication module, the on-board equipment of the following vehicles provides a prompt to the driver through the on-board human-machine interface.

[0015] Furthermore, preferably, the vehicle status information includes at least one of the following: windshield wiper status, vehicle speed, activation status of the vehicle stability control system, and headlight status.

[0016] Furthermore, preferably, the vehicle's video image information includes video image information of the road in front of the vehicle.

[0017] Furthermore, preferably, the early warning information includes the type of meteorological disaster, the disaster level, the geographical coordinates of the disaster location, the confidence level of the identification result, and the information generation timestamp.

[0018] Furthermore, preferably, the early warning system also employs a distance-direction-time triple filtering mechanism to filter the transmission of early warning information; the distance filtering refers to sending early warning information only to subsequent vehicles whose distance from the vehicle that issued the early warning information is within a preset threshold range; the direction filtering refers to sending early warning information only to subsequent vehicles whose angle with the driving direction of the vehicle that issued the early warning information is within a preset threshold range; the time filtering refers to subsequent vehicles, after receiving the early warning information, judging its timeliness based on the timestamp generated by the information, and not processing information that exceeds a preset valid time window.

[0019] Furthermore, preferably, the preset threshold for the distance is 2 kilometers, the preset threshold for the directional angle is 15 degrees, and the preset value for the effective time window is 30 seconds.

[0020] Furthermore, preferably, multiple vehicles equipped with the early warning system communicate with each other through the C-V2X communication module to form a distributed meteorological monitoring network.

[0021] Furthermore, preferably, the early warning steps are as follows:

[0022] Step 1, Multimodal Data Acquisition: During vehicle operation, the on-board edge computing module acquires video image information and vehicle status information of the vehicle, including the status of the windshield wipers, vehicle speed, activation status of the vehicle stability control system, and headlight status.

[0023] Step 2, Data Fusion and Disaster Identification: The edge computing module first performs multimodal perception fusion on the collected video image information and vehicle status information to form a dataset containing rich spatiotemporal features; then, the built-in embedded artificial intelligence model performs in-depth analysis and processing on the fused dataset to identify whether there are meteorological disasters in the current driving environment and determine their specific types and severity levels; the meteorological disaster types include heavy fog, rainstorms, snow accumulation, and road icing;

[0024] Step 3, Warning Information Generation and Broadcasting: Once a meteorological disaster reaching the warning level is identified, the edge computing module will immediately generate a structured warning information data packet based on the identification results. This warning information data packet includes at least: the meteorological disaster type, disaster level, geographical coordinates of the disaster location, confidence level of the identification results, and information generation timestamp. Subsequently, the warning system broadcasts the warning information data packet to the surrounding area via the C-V2X communication module in a direct vehicle-to-vehicle communication manner.

[0025] Step 4, Warning Information Filtering and Reception: Within the communication range of the vehicle sending the warning information, the onboard equipment of other subsequent vehicles will receive the warning information. The receiving vehicle's system does not immediately process all information, but first uses a distance-direction-time triple filtering mechanism to determine the validity of the information. The warning information is only deemed valid when the receiving vehicle simultaneously meets the following three conditions:

[0026] Condition 1: The distance to the originating vehicle is within a preset threshold;

[0027] Condition 2: The direction of travel is basically the same as that of the vehicle that issued the order;

[0028] Condition 3: The timestamp of the warning information has not expired;

[0029] Step 5, Warning Execution and Prompt: For valid warning information that has passed the triple filtering mechanism, the receiving vehicle's warning system will immediately issue a clear warning prompt to the driver through the in-vehicle human-machine interface; the prompt includes the type of disaster ahead and suggested driving measures.

[0030] Furthermore, preferably, the specific method of the distance-direction-time triple filtering mechanism in step four is as follows:

[0031] Assume the vehicle issuing the warning is v. i The subsequent vehicles that receive the warning information are v j Vehicle v j In T current Received from v at any time i The issued warning information data packet M warn Among them, M warn Includes the location of the dispatching vehicle. direction and information generation timestamp T gen ;

[0032] Vehicle v j Its own position is Direction is

[0033] Vehicle v jThe system will determine the validity of the information using the following three Boolean functions:

[0034] (1) Distance filtering function F dist

[0035] This function determines whether the distance between the two vehicles is within the effective transfer radius R. eff Inside;

[0036]

[0037] Where distance(L1,L2) is a function that calculates the distance between two geographic coordinate points; only when F dist Information is filtered by distance only when the condition is true.

[0038] (2) Directional filtering function F dir

[0039] This function determines whether the two vehicles are traveling in essentially the same direction, that is, whether the angle between their directions is within a preset threshold θ. th Inside;

[0040]

[0041] Where angle(D1,D2) is a function that calculates the angle between two direction vectors; it only applies when F dir Information is filtered by direction only when it is true.

[0042] (3) Time filtering function F time

[0043] This function determines whether the received information is still within the valid time window T. valid Inside;

[0044] F time =((T) current -T gen )≤T valid )

[0045] Only when F time Only when the time filter is true can the information pass through the time filter.

[0046] Final validity determination: Warning message M is only valid if the results of the above three filtering functions are all true. warn Only then was it determined to be a violation of vehicle v j It is effective and triggers subsequent prompts on the in-vehicle human-machine interface; the logical relationship is as follows:

[0047] isValid(M warn ) = F dist ∧F dir ∧F time

[0048] Where ∧ represents the logical AND operation; if isValid(M warn If ) is false, then vehicle v j The system will discard the warning message directly, without interfering with the driver.

[0049] This invention provides a meteorological disaster early warning system based on C-V2X and edge computing. The system includes: an onboard device installed in a vehicle, the onboard device integrating an edge computing module and an embedded artificial intelligence model; the onboard device collecting video image information and vehicle status information, and fusing the video image information and vehicle status information through multimodal perception; the embedded artificial intelligence model identifying meteorological disasters from the fused data; direct communication between vehicles based on C-V2X technology; and the transmission of meteorological disaster early warning information to subsequent vehicles traveling on the same route in the same direction, reminding drivers to drive cautiously or take necessary precautions in advance, thereby improving road traffic safety. Optionally, the system employs a triple filtering mechanism of "distance-direction-time" to screen early warning information and utilizes mobile vehicles to construct a distributed meteorological monitoring network.

[0050] Compared to CN 113727304A, an emergency vehicle early warning system and method based on a 5G communication architecture, this invention differs fundamentally in its technical problems, system architecture, and implementation. That technology discloses an emergency vehicle early warning system aimed at identifying and warning of special vehicles, which is completely different from the technical purpose of this invention: monitoring and warning of traffic and meteorological disasters. More importantly, the system architecture disclosed in that technology relies on roadside facilities, and its perception fusion processing is completed in a 5G mobile edge computing unit on the roadside, which then sends the processing results to the vehicle. In contrast, the computation and identification subject of this invention is the vehicle itself; the onboard equipment integrates an edge computing module, does not rely on specific roadside facilities, and has wider applicability. Furthermore, that technology does not disclose a scheme for achieving direct communication between vehicles using C-V2X technology. Therefore, that technology has inherent drawbacks such as reliance on roadside facilities, high deployment costs, and limited applicable scenarios, while this invention overcomes these limitations through a purely onboard edge computing solution, exhibiting greater universality and economy.

[0051] Compared to CN 106781697A, a method for real-time perception and collision avoidance warning of vehicle-mounted adverse weather, the ultimate technical goal and implementation path of this invention are fundamentally different. That technology discloses a method for real-time perception and collision avoidance warning of vehicle-mounted adverse weather, the core of which lies in using the perceived weather information as input parameters for a collision avoidance warning module to improve the collision avoidance performance of a single vehicle in dangerous situations, and ultimately achieve control over the vehicle. However, the purpose of this invention is not vehicle collision avoidance control, but rather to use the vehicle as a mobile monitoring node to construct a collaborative disaster information early warning network using C-V2X technology. To achieve this goal, this invention proposes a "distance-direction-time" triple filtering mechanism for warning information to achieve accurate information distribution, which is completely undisclosed in that technology. Therefore, it is evident that that technology only focuses on improving the intelligence of a single vehicle, failing to solve the problem of collaborative perception and early warning between vehicles, and unable to inform vehicles behind of potential dangers ahead. This invention, by constructing a C-V2X collaborative early warning network and creating a unique triple filtering mechanism, precisely makes up for this key deficiency, achieving a leap from "single vehicle protection" to "fleet collaborative safety".

[0052] Compared to CN 119169769A, a road traffic natural disaster early warning and prevention system based on AIoT, the system architecture of this invention is fundamentally different. That technology discloses an AIoT-based road traffic natural disaster early warning and prevention system, whose system architecture is based on fixed road infrastructure. Its edge layer module consists of various fixed sensors and intelligent natural disaster monitoring terminals deployed along the road. In contrast, this invention discloses a vehicle-centric mobile monitoring and early warning system. Its onboard equipment integrates edge computing and artificial intelligence models, using moving vehicles as nodes for perception, computation, and communication. Therefore, the fixed, roadside facility-based monitoring scheme disclosed in that technology is completely different in system composition and implementation from the mobile, vehicle-centric scheme proposed in this invention. The fixed facility-based scheme of that technology suffers from fundamental drawbacks such as high cost and blind spots in monitoring. This invention, however, transforms each vehicle into a mobile monitoring node, achieving dynamic and seamless coverage of the road environment at low marginal cost, thereby overcoming these drawbacks.

[0053] Compared with the prior art, the beneficial effects of this invention are as follows:

[0054] (1) It has strong real-time performance, and uses the vehicle edge computing module to analyze the weather conditions of the road ahead in real time;

[0055] (2) It has a wide coverage area and can transmit early warning information over a wide range through direct communication between vehicles; it can improve safety, provide early warning of severe weather conditions, and reduce the occurrence of traffic accidents.

[0056] (3) Low maintenance cost, the on-board equipment is located inside the vehicle, the operation stability is high, and maintenance is convenient;

[0057] (4) It is highly adaptable and the system can dynamically adjust the warning level according to the actual identification results; it is highly scalable and the system architecture supports the future integration of more sensor data.

[0058] This invention proposes a vehicle-mounted edge computing system for real-time monitoring and early warning of traffic and meteorological disasters using C-V2X technology. Utilizing vehicle-mounted edge computing and artificial intelligence, it achieves real-time monitoring and early warning of road weather conditions, improving road traffic safety and efficiency, and has broad application prospects. Through formal mathematical models and system architecture design, this invention provides an innovative solution for the field of traffic and meteorological monitoring, filling the gaps in traditional fixed observation station monitoring systems and offering a new technological path for the development of intelligent transportation systems. Attached Figure Description

[0059] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0060] Figure 2 This is a flowchart of the system workflow of the present invention;

[0061] Figure 3 This is a schematic diagram of the "distance-direction-time" triple filtering mechanism for early warning information in this invention. Detailed Implementation

[0062] The present invention will now be described in further detail with reference to the embodiments.

[0063] Those skilled in the art will understand that the following embodiments are for illustrative purposes only and should not be construed as limiting the scope of the invention. Where specific techniques or conditions are not specified in the embodiments, they are performed in accordance with the techniques or conditions described in the literature in the field or according to the product instructions. Materials or equipment whose manufacturers are not specified are all conventional products that can be obtained by purchase.

[0064] Example 1

[0065] A meteorological disaster early warning system based on C-V2X and edge computing includes vehicle-mounted equipment installed on a vehicle; the vehicle-mounted equipment includes an edge computing module and a C-V2X communication module.

[0066] The edge computing module is used to collect video image information and vehicle status information of the vehicle, perform multimodal perception fusion on the collected video image information and vehicle status information, and process the fused data through the embedded artificial intelligence model built into the edge computing module to identify the type and level of meteorological disasters in the vehicle's driving environment and generate corresponding early warning information.

[0067] The C-V2X communication module, connected to the edge computing module, is used to send warning information to following vehicles traveling in the same direction on the same route through direct inter-vehicle communication. After receiving the warning information through its C-V2X communication module, the on-board equipment of the following vehicles provides a prompt to the driver through the on-board human-machine interface.

[0068] Example 2

[0069] A meteorological disaster early warning system based on C-V2X and edge computing includes vehicle-mounted equipment installed on a vehicle; the vehicle-mounted equipment includes an edge computing module and a C-V2X communication module.

[0070] The edge computing module is used to collect video image information and vehicle status information of the vehicle, perform multimodal perception fusion on the collected video image information and vehicle status information, and process the fused data through the embedded artificial intelligence model built into the edge computing module to identify the type and level of meteorological disasters in the vehicle's driving environment and generate corresponding early warning information.

[0071] The C-V2X communication module, connected to the edge computing module, is used to send warning information to following vehicles traveling in the same direction on the same route through direct inter-vehicle communication. After receiving the warning information through its C-V2X communication module, the on-board equipment of the following vehicles provides a prompt to the driver through the on-board human-machine interface.

[0072] Vehicle status information includes at least one of the following: wiper status, vehicle speed, activation status of the vehicle stability control system, and headlight status.

[0073] The vehicle's video image information includes video image information of the road in front of the vehicle.

[0074] The early warning information includes the type of meteorological disaster, the disaster level, the geographical coordinates of the disaster location, the confidence level of the identification results, and the timestamp of information generation.

[0075] The early warning system also employs a triple filtering mechanism of distance, direction, and time to filter the transmission of early warning information. Distance filtering means that early warning information is only sent to subsequent vehicles whose distance from the vehicle that issued the early warning information is within a preset threshold range. Direction filtering means that early warning information is only sent to subsequent vehicles whose angle with the direction of travel of the vehicle that issued the early warning information is within a preset threshold range. Time filtering means that after receiving the early warning information, subsequent vehicles determine its timeliness based on the timestamp generated by the information and do not process information that exceeds a preset effective time window.

[0076] The preset threshold for the distance is 2 kilometers, the preset threshold for the directional angle is 15 degrees, and the preset value for the effective time window is 30 seconds.

[0077] Multiple vehicles equipped with early warning systems communicate with each other through the C-V2X communication module, forming a distributed meteorological monitoring network.

[0078] The warning steps are as follows:

[0079] Step 1, Multimodal Data Acquisition: During vehicle operation, the on-board edge computing module acquires video image information and vehicle status information of the vehicle, including the status of the windshield wipers, vehicle speed, activation status of the vehicle stability control system, and headlight status.

[0080] Step 2, Data Fusion and Disaster Identification: The edge computing module first performs multimodal perception fusion on the collected video image information and vehicle status information to form a dataset containing rich spatiotemporal features; then, the built-in embedded artificial intelligence model performs in-depth analysis and processing on the fused dataset to identify whether there are meteorological disasters in the current driving environment and determine their specific types and severity levels; the meteorological disaster types include heavy fog, rainstorms, snow accumulation, and road icing;

[0081] Step 3, Warning Information Generation and Broadcasting: Once a meteorological disaster reaching the warning level is identified, the edge computing module will immediately generate a structured warning information data packet based on the identification results. This warning information data packet includes at least: the meteorological disaster type, disaster level, geographical coordinates of the disaster location, confidence level of the identification results, and information generation timestamp. Subsequently, the warning system broadcasts the warning information data packet to the surrounding area via the C-V2X communication module in a direct vehicle-to-vehicle communication manner.

[0082] Step 4, Warning Information Filtering and Reception: Within the communication range of the vehicle sending the warning information, the onboard equipment of other subsequent vehicles will receive the warning information. The receiving vehicle's system does not immediately process all information, but first uses a distance-direction-time triple filtering mechanism to determine the validity of the information. The warning information is only deemed valid when the receiving vehicle simultaneously meets the following three conditions:

[0083] Condition 1: The distance to the originating vehicle is within a preset threshold;

[0084] Condition 2: The direction of travel is basically the same as that of the vehicle that issued the order;

[0085] Condition 3: The timestamp of the warning information has not expired;

[0086] Step 5, Warning Execution and Prompt: For valid warning information that has passed the triple filtering mechanism, the receiving vehicle's warning system will immediately issue a clear warning prompt to the driver through the in-vehicle human-machine interface; the prompt includes the type of disaster ahead and suggested driving measures.

[0087] In step four, the specific method of the distance-direction-time triple filtering mechanism is as follows:

[0088] Assume the vehicle issuing the warning is v. i The subsequent vehicles that receive the warning information are v j Vehicle v j In T current Received from v at any time i The issued warning information data packet M warn Among them, M warn Includes the location of the dispatching vehicle. direction and information generation timestamp T gen ;

[0089] Vehicle v j Its own position is Direction is

[0090] Vehicle v j The system will determine the validity of the information using the following three Boolean functions:

[0091] (1) Distance filtering function F dist

[0092] This function determines whether the distance between the two vehicles is within the effective transfer radius R. eff Inside;

[0093]

[0094] Where distance(L1,L2) is a function that calculates the distance between two geographic coordinate points; only when F dist Information is filtered by distance only when the condition is true.

[0095] (2) Directional filtering function F dir

[0096] This function determines whether the two vehicles are traveling in essentially the same direction, that is, whether the angle between their directions is within a preset threshold θ. th Inside;

[0097]

[0098] Where angle(D1,D2) is a function that calculates the angle between two direction vectors; it only applies when F dir Information is filtered by direction only when it is true.

[0099] (3) Time filtering function F time

[0100] This function determines whether the received information is still within the valid time window T. valid Inside;

[0101] F time =((T) current -T gen )≤T valid )

[0102] Only when F time Only when the time filter is true can the information pass through the time filter.

[0103] Final validity determination: Warning message M is only valid if the results of the above three filtering functions are all true. warn Only then was it determined to be a violation of vehicle v j It is effective and triggers subsequent prompts on the in-vehicle human-machine interface; the logical relationship is as follows:

[0104] isValid(M warn ) = F dist ∧F dir ∧F time

[0105] Where ∧ represents the logical AND operation; if isValid(M warn If ) is false, then vehicle v j The system will discard the warning message directly, without interfering with the driver.

[0106] Example 3

[0107] A meteorological disaster early warning system based on C-V2X and edge computing includes vehicle-mounted equipment installed on a vehicle; the vehicle-mounted equipment includes an edge computing module and a C-V2X communication module.

[0108] The edge computing module is used to collect video image information and vehicle status information of the vehicle, perform multimodal perception fusion on the collected video image information and vehicle status information, and process the fused data through the embedded artificial intelligence model built into the edge computing module to identify the type and level of meteorological disasters in the vehicle's driving environment and generate corresponding early warning information.

[0109] The C-V2X communication module, connected to the edge computing module, is used to send warning information to following vehicles traveling in the same direction on the same route through direct inter-vehicle communication. After receiving the warning information through its C-V2X communication module, the on-board equipment of the following vehicles provides a prompt to the driver through the on-board human-machine interface.

[0110] A meteorological disaster early warning method based on C-V2X and edge computing is proposed, employing a meteorological disaster early warning system based on C-V2X and edge computing, including the following steps:

[0111] Step 1, Multimodal Data Acquisition: During vehicle operation, the edge computing module within the onboard equipment collects two types of data in real time and synchronously: the first type is video image information of the road ahead of the vehicle captured by the front-facing camera; the second type is vehicle status information from the vehicle controller area network (CAN) bus or related sensors, specifically including wiper status, vehicle speed, activation status of the vehicle stability control system (e.g., ESP), and headlight status.

[0112] Step 2, Data Fusion and Disaster Identification: The edge computing module first performs multimodal perception fusion on the collected video image information and vehicle status information to form a dataset containing rich spatiotemporal features; then, the built-in embedded artificial intelligence model performs in-depth analysis and processing on the fused dataset to identify whether there are meteorological disasters in the current driving environment and determine their specific types and severity levels; the meteorological disaster types include heavy fog, rainstorms, snow accumulation, and road icing;

[0113] Step 3, Warning Information Generation and Broadcasting: Once a meteorological disaster reaching the warning level is identified, the edge computing module will immediately generate a structured warning information data packet based on the identification results. This warning information data packet includes at least: the meteorological disaster type, disaster level, geographical coordinates of the disaster location, confidence level of the identification results, and information generation timestamp. Subsequently, the warning system broadcasts the warning information data packet to the surrounding area via the C-V2X communication module in a vehicle-to-vehicle (V2V) communication manner.

[0114] Step 4, Warning Information Filtering and Reception: Within the communication range of the vehicle sending the warning information, the onboard equipment of other subsequent vehicles will receive the warning information. The warning system of the receiving vehicle will not process all information immediately, but will first use the "distance-direction-time" triple filtering mechanism proposed in this invention to determine the validity of the information. The warning information is only deemed valid when the receiving vehicle simultaneously meets the following three conditions: (1. Distance condition) The distance to the sending vehicle is within a preset threshold; (2. Direction condition) The direction of travel of the receiving vehicle is basically consistent with that of the sending vehicle; (3. Time condition) The timestamp of the warning information has not expired.

[0115] Step 5, Warning Execution and Notification: For valid warning information that has passed the triple filtering mechanism, the receiving vehicle's warning system will immediately issue a clear warning notification to the driver through the in-vehicle human-machine interface, such as by popping up a warning window on the central control screen and highlighting the disaster-prone road section, and simultaneously by broadcasting a voice notification through the vehicle's audio system. The notification includes the type of disaster ahead and suggested driving measures, enabling the driver to take necessary measures in advance, such as slowing down, maintaining a safe distance, or planning an alternative route, to ensure driving safety.

[0116] (1) Formal System Expression

[0117] To clearly illustrate the core technical solution of this invention, the "distance-direction-time" triple filtering mechanism executed by subsequent vehicles after receiving the warning information is formally expressed. This mechanism is crucial for ensuring the accurate and effective delivery of warning information.

[0118] Assume the vehicle issuing the warning is v. i The subsequent vehicles that receive the warning information are v j Vehicle v j In T current Received from v at any time i The issued warning information data packet M warn .

[0119] Among them, M warn Includes the location L of the dispatching vehicle vi Direction D vi and information generation timestamp T gen ;

[0120] Vehicle v j Its own position is Direction is

[0121] Vehicle v j The system will use the following three Boolean functions to determine whether the information is valid:

[0122] 1. Distance filtering function (F) dist )

[0123] This function determines whether the distance between the two vehicles is within the effective transfer radius R. eff Inside. R eff Set preset parameters for the system, such as 2 kilometers.

[0124]

[0125] Here, distance(L1,L2) is a function that calculates the distance between two geographic coordinate points. Only when F... dist Information is filtered by distance only when the result is true.

[0126] 2. Directional filtering function (F) dir )

[0127] This function determines whether the two vehicles are traveling in essentially the same direction, that is, whether the angle between their directions is within a preset threshold θ. th Inside. θ th Set preset parameters for the system, such as 15 degrees.

[0128]

[0129] Here, angle(D1,D2) is a function that calculates the angle between two direction vectors. Only when F... dir Information is filtered by direction only when the result is true.

[0130] 3. Time filtering function (F) time )

[0131] This function determines whether the received information is still within the valid time window T. valid Internally, to ensure the timeliness of information. valid Set preset parameters for the system, such as 30 seconds.

[0132] F time =((T) current -T gen )≤T valid )

[0133] Only when F time Information is filtered through time only when the result is true.

[0134] Final validity ruling

[0135] Warning message M is only issued when the results of the above three filtering functions are all true. warn Only then was it determined to be a violation of vehicle v j It is effective and triggers subsequent prompts on the in-vehicle human-machine interface. The logical relationship is as follows:

[0136] isValid(M warn ) = F dist ∧F dir ∧F time

[0137] Where ∧ represents the logical AND operation. If isValid(M warn If ) is false, then vehicle v j The system will discard the warning message directly, without interfering with the driver.

[0138] (2) System Architecture

[0139] like Figure 1 As shown, the system includes an onboard edge computing module and a C-V2X communication module. The onboard edge computing module, installed in the vehicle, is responsible for real-time acquisition of video images of the road ahead and uses an embedded artificial intelligence model to identify weather conditions. The C-V2X communication module enables direct communication between vehicles, transmitting warning information between them.

[0140] (3) Training of artificial intelligence models

[0141] The embedded artificial intelligence model in this invention is trained on a large, labeled traffic and meteorological dataset using deep learning technology. The core capability of this model lies in its ability to effectively handle spatiotemporal features: it can not only analyze spatial features in single-frame video images to identify scene content, but also understand temporal features in video sequences and vehicle status time series to determine the dynamic evolution of meteorological conditions. The system framework of this invention does not limit the specific structure of the artificial intelligence model; any advanced artificial intelligence model capable of effectively extracting spatiotemporal features from the aforementioned multimodal fusion data to complete the meteorological disaster identification task is applicable to this invention. In this way, the system can guarantee high accuracy and robustness in disaster identification, while also possessing the ability to iteratively upgrade as artificial intelligence technology develops.

[0142] When the model is running, its input is a synchronized multimodal data stream, including a continuous sequence of video images representing the road environment, and a vehicle state time series that is precisely synchronized with it. The vehicle state information includes the status of the windshield wipers, vehicle speed, the activation status of the vehicle stability control system, and the status of the headlights.

[0143] After comprehensively analyzing the above inputs, the model outputs a refined identification result of specific physical meteorological conditions. For example, the model's output may include classification predictions of precipitation events (such as identifying "moderate rain," "heavy rain," or "extremely heavy rain"), numerical predictions of visibility (such as "180 meters"), and classification predictions of road surface conditions (such as identifying "slippery" or "icy"). The identification results output by the model and their corresponding prediction probabilities or evaluation scores constitute the identification results and their confidence levels described in this invention. When the confidence level of an identification result meets preset conditions, the system adopts that result and, according to the rules defined in the "Meteorological Disaster Classification Mechanism," matches the identified specific physical conditions to the corresponding warning level, thereby triggering subsequent warning procedures.

[0144] This invention does not limit the specific structure of the embedded artificial intelligence model; any model that can achieve the functions of this invention is acceptable. For example, the following types of models can be used:

[0145] (1) Two-dimensional convolutional neural networks and lightweight network models, such as ResNet, EfficientNet, MobileNet, ShuffleNet, and GhostNet;

[0146] (2) Temporal modeling network models, such as 3D CNNs: C3D, I3D, X3D, R(2+1)D; ConvLSTM; TCN;

[0147] (3) Visual or video model based on Transformer, such as spatiotemporal attention networks like ViT, TimeSformer, VideoSwin, and UniFormer.

[0148] (4) Cross-modal fusion architecture models, such as cross-attention multimodal Transformer, gated fusion, and weighted mid-to-late stage fusion;

[0149] (5) Auxiliary feature extraction module models, such as the YOLO series of target detection / segmentation networks, CenterNet, DeepLabv3+, and U-Net, are used for explicit extraction of road and weather elements; optical flow / motion estimation networks are used for dynamic feature enhancement.

[0150] To adapt to in-vehicle edge computing resources, the model can be selected using model compression and acceleration methods such as quantization, pruning, and distillation, without changing the technical essence of the present invention.

[0151] (4) Multimodal data fusion

[0152] A key technical feature of this invention is that the system not only utilizes video images for analysis but also innovatively integrates the vehicle's own state information for multimodal perception, thereby significantly improving the accuracy and robustness of meteorological disaster identification. The edge computing module uses both video image information and vehicle state information as input to the embedded artificial intelligence model. For example, when judging precipitation intensity, if the artificial intelligence model initially identifies "rainy" weather through video images, the system will further read the windshield wiper status. If the wipers are in a high-speed wiping state, the disaster level will be upgraded to "heavy rain" or the corresponding recognition confidence level will be increased. When judging road slipperiness or icing conditions, the system integrates the activation status of the vehicle stability control system (ESP, Electronic Stability Program). Here, "activation status" refers to the ESP system actively intervening to prevent wheel slippage; this status can be directly obtained from the vehicle's CAN bus. In low-temperature environments, if the ESP is activated and the video image shows a wet road surface, the system can determine with high probability that there is a risk of icing on the road surface, even if the human eye or camera cannot directly distinguish thin ice. Similarly, when assessing visibility, if the model identifies "fog" or "rain," and the system detects that the driver has turned on the headlights during the day, this driving behavior can serve as corroborating evidence to confirm that the current visibility does indeed affect driving safety, and the warning level will be raised accordingly. Through the above-mentioned multimodal data fusion, this invention combines objective visual information with behavioral and status information that reflects the driver's subjective judgment and the vehicle's actual dynamics, establishing a more reliable disaster identification model.

[0153] (5) Meteorological disaster classification mechanism

[0154] This invention establishes a specific meteorological disaster graded early warning mechanism. The system can automatically grade the severity of disasters and trigger corresponding warnings based on the identification results of an artificial intelligence model using multimodal data fusion analysis. This mechanism sets clear warning thresholds for different meteorological disasters, mainly divided into three levels: Level 1 warning (yellow) corresponds to situations that slightly affect driving safety, triggered when the artificial intelligence model identifies moderate rain or snow, or dense fog with visibility greater than 200 meters and less than or equal to 500 meters; Level 2 warning (orange) corresponds to situations that significantly affect driving safety, triggered when the system identifies heavy rain or snow, or dense fog with visibility decreasing to greater than 50 meters and less than or equal to 200 meters. Additionally, a Level 2 warning is also triggered when multimodal fusion analysis determines the current road surface is slippery; Level 3 warning (red) corresponds to extreme situations that severely affect driving safety, triggered when the system identifies extremely heavy rain or snowstorms, or extremely dense fog with visibility less than or equal to 50 meters, or when the vehicle stability control system is activated and it is clearly determined that the road surface is icy. Different levels of warnings will trigger different driving advice. Level 1 warning (yellow) triggers the advice to "drive cautiously and reduce speed appropriately". Level 2 warning (orange) triggers the advice to "reduce speed immediately and maintain a safe distance". Level 3 warning (red) triggers the highest level advice to "extreme danger, it is recommended to find a safe place to stop", thus achieving refined safety prompts.

[0155] (6) Issuance of early warning information

[0156] When the onboard edge computing module detects severe weather conditions ahead that reach the warning level, the system immediately generates a structured warning information data packet and broadcasts it to the surrounding area via C-V2X V2V communication. This warning information data packet specifically includes the following: the geographic coordinates of the disaster location, the identified type of meteorological disaster, the severity level of the disaster, the confidence level of the AI ​​model's identification, the information generation timestamp, and automatically matched suggested driving measures based on the disaster level. It should be noted that while this information is physically a localized broadcast, the precise delivery to "following vehicles traveling on the same route and in the same direction" is ultimately achieved through a "distance-direction-time" triple filtering mechanism executed by the receiving vehicle, thus ensuring that vehicles in oncoming lanes or traveling on unrelated routes are not affected by this warning information.

[0157] (7) Early warning information transmission mechanism

[0158] To ensure accurate delivery of early warning information, this invention innovatively designs a triple filtering mechanism of "distance-direction-time" executed by the receiving vehicle. The principle is as follows: Figure 3As shown. This mechanism does not involve filtering before sending; instead, the receiving vehicle performs a strict autonomous judgment on all local warning information received via C-V2X according to the following rules: First, distance filtering is performed. The receiving vehicle's system calculates the geographical distance between itself and the disaster location indicated in the warning information. Only when this distance is less than a preset effective transmission radius (e.g., 2 kilometers) does the information proceed to the next stage. Second, direction filtering is performed. The system compares its current driving direction with the driving direction of the sending vehicle contained in the warning information. Only when the angle between the two is less than a preset direction difference threshold (e.g., 15 degrees) is the direction considered relevant. Finally, time filtering is performed. The system compares the current time with the timestamp of the warning information. Only when the time difference is less than a preset effective time window (e.g., 30 seconds) is the timeliness of the information confirmed. Only when a warning information passes all three filters is it ultimately determined to be valid. For valid warning information, the system will immediately issue a warning through the in-vehicle human-machine interface, such as displaying a graphic warning on the central control screen, highlighting the disaster section on the navigation map, and providing a voice broadcast in conjunction with the in-vehicle audio system. Any information that fails to pass any filtering stage will be silently discarded by the system in the background, thus achieving accurate delivery of warning information and avoiding invalid information interference to drivers.

[0159] (8) Workflow and Status Transition

[0160] like Figure 2 As shown, the system's workflow includes an initial state, image acquisition state, meteorological identification state, decision-making state, early warning issuance state, and early warning continuation state. After the system starts and enters the idle monitoring state, it periodically acquires and analyzes images. Based on the analysis results, it determines whether to issue an early warning. If the identification result has a high confidence level and the meteorological level falls within the warning range, it sends an early warning message and updates it periodically until the disaster area is left or the meteorological conditions improve.

[0161] Application Examples

[0162] To further illustrate the technical effects of the present invention, a specific application example is provided below.

[0163] 1. Scene setting

[0164] Location: A mountain highway. Due to the terrain, small-scale, sudden patches of fog often appear at K10+500, with visibility dropping sharply to below 50 meters.

[0165] Current technology deployment: There is a traditional roadside device at K5 and K15, which cannot cover the fog area at K10+500.

[0166] vehicle:

[0167] Vehicle A: The first vehicle about to enter the foggy area is equipped with the system of this invention.

[0168] Vehicle B: A vehicle traveling in the same direction, approximately 800 meters behind Vehicle A, also equipped with the system of this invention.

[0169] Vehicle C: A vehicle traveling in the opposite lane, also equipped with the system of this invention.

[0170] Initial conditions: All vehicles are traveling normally at a speed of 100 km / h, the weather is clear, and visibility is good.

[0171] 2. Comparison of Technical Solutions

[0172] (1) Implementation process of the present invention:

[0173] 1. Disaster Identification and Early Warning Generation: At time T0, vehicle A enters the dense fog area at K10+500. Its onboard equipment immediately collects the feature of a sudden drop in visibility in the video image, and combines it with vehicle status information (driver turns on fog lights). The built-in artificial intelligence model identifies "Level 3 (Red) Warning: Dense Fog" within 200 milliseconds, and generates a structured early warning data package containing information such as disaster type, location (K10+500), and timestamp.

[0174] 2. Information Broadcasting and Filtering: At time T0+0.25, vehicle A broadcasts the warning information via the C-V2X communication module. Vehicles B and C receive this information simultaneously.

[0175] Vehicle B's handling: The system immediately activates a triple filtering mechanism of "distance-direction-time". It is determined that Vehicle B is 800 meters away from Vehicle A (less than the preset 2-kilometer threshold), traveling in the same direction (angle less than 15 degrees), and the information timestamp is valid. Therefore, the warning is deemed valid.

[0176] Handling of Vehicle C: During the filtering process, the system detects that its travel direction is opposite to that of Vehicle A (angle greater than 15 degrees), thus failing to meet the direction filtering criteria. Therefore, the system discards this information directly without interfering with the driver.

[0177] 3. Warning Execution and Effect: At T0+0.3 seconds, the driver of vehicle B received a clear warning on the central control screen and through voice prompts: "Attention, you are entering a dense fog section 800 meters ahead. Visibility is less than 50 meters. Please slow down immediately!" The driver had sufficient reaction time (approximately 28 seconds) to calmly reduce the vehicle speed to a safe range and smoothly enter the dense fog area, effectively avoiding the risk of rear-end collisions caused by sudden braking due to severe weather.

[0178] (2) Comparison process using existing technologies:

[0179] 1. Monitoring and Dissemination: Because the fog occurred in the blind spot between the two roadside monitoring devices, neither the K5 nor K15 roadside monitoring devices detected any abnormalities. Therefore, the traffic management center was unable to obtain disaster information for this area, let alone issue a warning.

[0180] 2. Driving Situation and Consequences: Vehicle A slowed down suddenly after entering dense fog. The driver of vehicle B, unable to see any landmarks and anticipate the danger ahead, continued to approach at a high speed of 100 km / h. By the time the driver of vehicle B could see the taillights of vehicle A, the distance was already extremely close, forcing them to brake suddenly, which could easily have caused a rear-end collision or even a serious multi-vehicle pileup.

[0181] 3. Analysis of Beneficial Effects

[0182] As can be seen from the comparison of the above embodiments, the present invention has the following outstanding advantages compared with the prior art:

[0183] (1) Real-time and forward-looking: This invention achieves sub-second "discovery and warning", and the delay in issuing warning information is much lower than the several minutes delay of existing roadside equipment solutions, providing valuable prediction and reaction time for subsequent drivers.

[0184] (2) Coverage accuracy and breadth: This invention turns every vehicle into a mobile monitoring post, solving the problem of monitoring blind spots caused by the sparse deployment of traditional roadside equipment, and achieving “seamless” and refined coverage of the road environment.

[0185] (3) Information accuracy: The unique “distance-direction-time” triple filtering mechanism ensures that the warning information is delivered only to the most relevant vehicles, effectively avoiding interference from irrelevant information, solving the “alarm fatigue” problem caused by traditional broadcasting methods, and greatly improving the effectiveness and credibility of the warning.

[0186] The foregoing has shown and described the basic principles, main features, and technical advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A meteorological disaster early warning system based on C-V2X and edge computing, characterized in that, This includes onboard equipment installed on the vehicle; the onboard equipment includes an edge computing module and a C-V2X communication module; The edge computing module is used to collect video image information and vehicle status information of the vehicle, perform multimodal perception fusion on the collected video image information and vehicle status information, and process the fused data through the embedded artificial intelligence model built into the edge computing module to identify the type and level of meteorological disasters in the vehicle's driving environment and generate corresponding early warning information. The C-V2X communication module, connected to the edge computing module, is used to send warning information to following vehicles traveling in the same direction on the same route through direct inter-vehicle communication. After receiving the warning information through its C-V2X communication module, the on-board equipment of the following vehicles provides a prompt to the driver through the on-board human-machine interface.

2. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, Vehicle status information includes at least one of the following: wiper status, vehicle speed, activation status of the vehicle stability control system, and headlight status.

3. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, The vehicle's video image information includes video image information of the road in front of the vehicle.

4. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, The early warning information includes the type of meteorological disaster, the disaster level, the geographical coordinates of the disaster location, the confidence level of the identification results, and the timestamp of information generation.

5. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, The early warning system also employs a triple filtering mechanism of distance, direction, and time to filter the transmission of early warning information. Distance filtering means that early warning information is only sent to subsequent vehicles whose distance from the vehicle that issued the early warning information is within a preset threshold range. Direction filtering means that early warning information is only sent to subsequent vehicles whose angle with the direction of travel of the vehicle that issued the early warning information is within a preset threshold range. Time filtering means that after receiving the early warning information, subsequent vehicles determine its timeliness based on the timestamp generated by the information and do not process information that exceeds a preset effective time window.

6. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, The preset threshold for the distance is 2 kilometers, the preset threshold for the directional angle is 15 degrees, and the preset value for the effective time window is 30 seconds.

7. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, Multiple vehicles equipped with early warning systems communicate with each other through the C-V2X communication module, forming a distributed meteorological monitoring network.

8. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 1, characterized in that, The warning steps are as follows: Step 1, Multimodal Data Acquisition: During vehicle operation, the on-board edge computing module acquires video image information and vehicle status information of the vehicle, including the status of the windshield wipers, vehicle speed, activation status of the vehicle stability control system, and headlight status. Step 2, Data Fusion and Disaster Identification: The edge computing module first performs multimodal perception fusion on the collected video image information and vehicle status information to form a dataset containing rich spatiotemporal features; Subsequently, the built-in embedded artificial intelligence model performs in-depth analysis and processing on the fused dataset to identify whether there are meteorological disasters in the current driving environment and to determine their specific types and severity levels; the meteorological disaster types include heavy fog, rainstorms, snow accumulation, and road icing; Step 3, Early Warning Information Generation and Broadcast: Once a meteorological disaster reaching the early warning level is identified, the edge computing module will immediately generate a structured early warning information data packet based on the identification results. This early warning information data packet will include at least: meteorological disaster type, disaster level, geographic coordinates of the disaster location, confidence level of the identification results, and information generation timestamp. Subsequently, the early warning system broadcasts the early warning information data packet to the surrounding area via the C-V2X communication module in a direct vehicle-to-vehicle communication manner; Step 4, Warning Information Filtering and Reception: Within the communication range of the vehicle sending the warning information, the onboard equipment of other subsequent vehicles will receive the warning information. The receiving vehicle's system does not immediately process all information, but first uses a distance-direction-time triple filtering mechanism to determine the validity of the information. The warning information is only deemed valid when the receiving vehicle simultaneously meets the following three conditions: Condition 1: The distance to the originating vehicle is within a preset threshold; Condition 2: The direction of travel is basically the same as that of the vehicle that issued the order; Condition 3: The timestamp of the warning information has not expired; Step 5, Warning Execution and Notification: For valid warning information that has passed the triple filtering mechanism, the receiving vehicle's warning system will immediately issue a clear warning notification to the driver through the in-vehicle human-machine interface; The information includes the types of hazards ahead and recommended driving actions.

9. The meteorological disaster early warning system based on C-V2X and edge computing according to claim 8, characterized in that, In step four, the specific method of the distance-direction-time triple filtering mechanism is as follows: Assume the vehicle issuing the warning is v. i The subsequent vehicles that receive the warning information are v j Vehicle v j In T current Received from v at any time i The issued warning information data packet M warn Among them, M warn Includes the location of the dispatching vehicle. direction and information generation timestamp T gen ; Vehicle v j Its own position is Direction is Vehicle v j The system will determine the validity of the information using the following three Boolean functions: (1) Distance filtering function F dist This function determines whether the distance between the two vehicles is within the effective transfer radius R. eff Inside; Where distance(L1,L2) is a function that calculates the distance between two geographic coordinate points; only when F dist Information is filtered by distance only when the condition is true. (2) Directional filtering function F dir This function determines whether the two vehicles are traveling in essentially the same direction, that is, whether the angle between their directions is within a preset threshold θ. th Inside; Where angle(D1,D2) is a function that calculates the angle between two direction vectors; it only applies when F dir Information is filtered by direction only when it is true. (3) Time filtering function F time This function determines whether the received information is still within the valid time window T. valid Inside; F time =((T current -T gen )≤T valid ) Only when F time Only when the time filter is true can the information pass through the time filter. Final validity determination: Warning message M is only valid if the results of the above three filtering functions are all true. warn Only then was it determined to be a violation of vehicle v j It is effective and triggers subsequent prompts on the in-vehicle human-machine interface; the logical relationship is as follows: isValid(M warn )=F dist ∧F dir ∧F time Where ∧ represents the logical AND operation; if isValid(M warn If ) is false, then vehicle v j The system will discard the warning message directly, without interfering with the driver.

Images