A multi-modal fusion early warning information distribution method and system based on expressway sudden weather disasters

Abstract

The application discloses a kind of multi-modal fusion early warning information distribution method and system under expressway sudden weather disaster, it is related to intelligent transportation and emergency communication technical field, including: the mechanism of sudden weather disaster multi-source perception and audience circle selection is constructed, audience target list is generated;Based on the intelligent identification and grading of terminal capability of TAC library, establish multi-modal early warning content adaptive adaptation engine, generate heterogeneous protocol data adapted to terminal type;Formulate multi-channel concurrent re-strategy with priority, the sending state of each channel is monitored in real time, and the target of sending failure is cross-channel to send again, to ensure that early warning information "zero omission";Through multi-channel concurrent, the early warning information distribution method of sudden weather disaster that realizes "must reach" effect improves the effectiveness and coverage of early warning, reduces traffic accident rate under adverse weather conditions.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and emergency communication technology, specifically a method and system for distributing multimodal fusion early warning information under sudden meteorological disasters on highways. Background Technology

[0002] Sudden meteorological disasters such as patchy fog, freezing rain, and strong crosswinds have a significant impact on traffic safety. Patchy fog, in particular, is characterized by its sudden onset, strong localization, and extremely low visibility, earning it the nickname "mobile killer" on highways. Current methods for issuing weather warnings mainly rely on standardized mass text messaging (SMS), radio broadcasts, or variable message signs (VMS) along roadsides. However, these methods have significant shortcomings: First, the "one-size-fits-all" approach is inefficient, failing to differentiate delivery based on the type of receiving terminal. For example, ordinary text messages are often ignored by drivers of connected vehicles, preventing direct intervention in the vehicle control system; while multimedia messages cannot be displayed for drivers using feature phones (especially older models). Second, the lack of an adaptive format conversion mechanism results in poor display of critical information on different terminals, sometimes even leading to garbled text or interception. Third, the lack of a feedback mechanism based on terminal capabilities makes it impossible to confirm whether the warning information has been effectively received after it is issued, hindering the formation of a closed loop.

[0003] While some existing improvement technologies propose location-based precise push notifications, most fail to fully utilize the communication network's ability to perceive user terminal capabilities, nor do they effectively integrate the complementary advantages of multiple modalities such as 5G messaging, vehicle-to-everything (V2X) signaling, and intelligent voice. Therefore, there is an urgent need to research a method for distributing sudden meteorological disaster warning information that can intelligently identify user terminal types, adaptively adapt warning content formats, and achieve a "guaranteed delivery" effect through multi-channel concurrency. This would improve the effectiveness and coverage of warnings and reduce traffic accident rates under severe weather conditions. Summary of the Invention

[0004] To address the shortcomings of existing early warning methods, such as limited versatility and poor terminal compatibility, this invention aims to provide a multimodal fusion early warning information distribution method and system for sudden meteorological disasters on highways. First, a multi-channel message center based on operator big data is constructed to obtain a real-time list of users within the disaster area. Then, the device capability database (TAC database) on the core network side is used to "profile" terminals within the area, identifying whether they are intelligent connected vehicle systems, 5G smart terminals, or ordinary functional terminals. Next, a multimodal early warning content adaptive adaptation engine is established to automatically convert the same meteorological early warning information into various formats such as control signaling, Rich Communication Services (RCS), Flash Messages (USSD), or Intelligent Voice Response (IVR). Finally, a hierarchical concurrent push strategy is implemented based on the early warning level and terminal type to achieve accurate delivery and effective warning of the information.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways includes: (1) Construct a multi-source perception and audience selection mechanism for sudden meteorological disasters By integrating meteorological bureau micro-monitoring data, highway roadside sensor data, and operator base station integrated sensing data, meteorological disaster information is obtained, and meteorological disasters that trigger early warnings are identified. Based on operator base station positioning technology, a dynamic electronic fence is established to select all mobile communication terminal users who are about to enter or are already in the disaster-affected area in real time, and a target audience list is generated. (2) Intelligent identification and classification of terminal capabilities based on TAC library By using the device identifier register or big data analysis platform of the operator's core network, the first 8 digits of the TAC of each IMEI in the audience target list are parsed, and the terminal capability database is queried to determine the type of the audience terminal; (3) Establish a multimodal early warning content adaptive adaptation engine Standardized meteorological warning metadata is pre-set, and heterogeneous protocol data adapted to the type of the target terminal is generated based on the type of the target terminal. (4) Perform priority-based multi-channel concurrent and blind-spot distribution. Develop a priority-based multi-channel concurrent retransmission strategy, monitor the transmission status of each channel in real time, and retransmit failed targets across channels to ensure that warning information is "zero omission".

[0006] In this invention, the process of determining the meteorological disaster that triggers the early warning is as follows: The weighted average method is used to calculate the comprehensive visibility index. : ; in, This is the normalized visibility value; The fog obstruction rate identified in the image; D 3 represents the attenuation characteristic value of the millimeter-wave channel; , and They are respectively , and D The weighting factor is 3; Set warning trigger threshold When the comprehensive visibility index Less than the warning trigger threshold When a disaster is detected on a section of road, a warning is triggered, and a unique disaster event ID is generated.

[0007] In this invention, the selection of mobile communication terminal users in step (1) is as follows: Using the location management function component of the operator's core network, a list of all users currently within the electronic fence is retrieved in real time. To eliminate nearby static users, the velocity vector of each user is calculated by combining base station handover signaling and Doppler frequency shift data. ; The selection logic is as follows: ①Condition A: The user is in a connected or idle state but has a location update within the fenced cell; ②Condition B: Set a movement speed threshold; the user's movement speed must be greater than the movement speed threshold. ③Condition C: The angle between the user's movement direction and the coordinates of the center of the fog patch. ; Users who meet all three of the above conditions are selected for IMSI / IMEI.

[0008] In this invention, the types of audience terminals include: The first type of terminal is the "intelligent connected vehicle unit", which has V2X communication and vehicle control capabilities; the second type of terminal is the "5G intelligent handheld terminal", which supports 5G messaging and high-definition video playback; the third type of terminal is the "basic function terminal", which only supports voice and ordinary text messages.

[0009] In this invention, for a type of terminal, the engine encapsulates metadata into V2X standard control signaling, including speed limit thresholds and braking commands; for a type of terminal, the engine automatically generates 5G rich media cards containing dynamic maps, on-site GIF animations, and text descriptions; for a type of terminal, the engine extracts core text information to generate flash message text that is forced to pop up or generates an emergency call script through TTS.

[0010] In this invention, the priority-based multi-channel concurrent reprocessing strategy and blind spot distribution in step (4) are as follows: Establish three sending queues and simultaneously call three gateway interfaces: ①V2X Gateway: Sends RSI signaling to a type of terminal via roadside RSU broadcast and Uu interface unicast; ②5G Messaging Center: Distributes rich media cards to Class II terminals via the IMS network; ③ Signaling gateway: Sends Flash SMS messages to three types of terminals; V2X signaling is prioritized for Class I terminals, while 5G messages are sent to Class II terminals bound to the vehicle. 5G rich media cards are prioritized for Class 1 terminals, while Class 2 terminals that have not received a confirmation receipt are automatically downgraded and triggered to make a strong reminder via intelligent voice call (IVR). For the three types of terminals, the flash message domination method is directly adopted.

[0011] In this invention, the blind-filling logic is as follows: ① Anchor point 1, cross-platform blind spot filling from vehicle infotainment system to mobile phone: After sending V2X signaling, a timer is started. If no application layer confirmation is received from the vehicle's infotainment system within the specified time, the system is determined to be offline or the OBU is faulty. The vehicle owner's mobile phone number bound to the infotainment system is immediately retrieved, the number is added to the "Class II Terminal" sending queue, and a 5G message is sent. ② Anchor point 2, cross-domain gap filling from data to speech: After sending a 5G message, the system starts a timer. If a status report is received indicating that the transmission failed or timed out, it usually means that the user is in a weak network environment. The interactive voice response interface of the core network is immediately triggered to initiate an automatic outbound call based on VoLTE. After the call is connected, the generated TTS voice script is played directly.

[0012] In this invention, the meteorological early warning metadata includes disaster type, location coordinates, suggested vehicle speed, evacuation route, and on-site video stream.

[0013] A multimodal fusion early warning information distribution system for sudden meteorological disasters on highways, characterized in that it includes: The perception layer includes several visibility meters and micro-weather stations deployed at intervals along the highway to collect temperature, humidity, and visibility data; high-definition AI cameras deployed at intervals along the highway, equipped with defogging algorithms and image recognition capabilities, to capture fog patterns and traffic density; and reuse of existing 5G-A base stations from operators' networks, utilizing their integrated communication and perception capabilities to detect object outlines in low visibility conditions through millimeter-wave radar echo characteristics. Network Layer: Based on the operator's 5G SA standalone network architecture, the access network is responsible for terminal connection; MEC edge computing nodes are deployed in the aggregation room closest to the highway, responsible for real-time data cleaning and low-latency decision-making; the core network provides user authentication, mobility management and signaling control. Platform layer: Deploy the core "Multimodal Fusion Early Warning Center", which includes three key subsystems: TAC terminal capability database, used to query device profiles; content adaptation engine, used for format conversion; and unified message distribution gateway, which integrates SMSC, MMSC, and V2X Server interfaces. Terminal layer: It covers all types of terminals currently in operation, from vehicle units that support the C-V2X protocol to 5G smartphones that support RCS, and traditional feature phones that only support 2G / 3G networks.

[0014] Compared with the prior art, the beneficial effects of the present invention are: 1. Based on operator base station positioning technology, this invention establishes dynamic electronic fences and real-time selection of all mobile communication terminal users who are about to enter or are already in the disaster-affected area, thereby achieving accurate identification of the target terminal type and accurate push of early warning content, changing the inefficient "blind" mode of the past.

[0015] 2. This invention fully and effectively integrates the complementary advantages of multiple modalities such as 5G messaging, vehicle-to-everything (V2X) signaling, and intelligent voice by adaptively adapting heterogeneous protocol data to multimodal content. It can be reached by different terminals in a differentiated manner, enabling intelligent vehicles to directly parse instructions and automatically decelerate, allowing smartphone users to intuitively view road conditions and detour in advance, and enabling feature phone users to receive warnings through voice or pop-up screens.

[0016] 3. This invention performs priority-based multi-channel concurrent and blind spot distribution, which significantly improves the reach and execution rate of early warning information under extreme weather conditions such as sudden fog, and effectively ensures traffic safety. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a diagram of the overall logical architecture of the present invention.

[0019] Figure 3 This is a schematic diagram of the logic for multi-source sensing data fusion and fog disaster determination in this invention.

[0020] Figure 4 This is a tree diagram for the intelligent identification and hierarchical determination of terminal capabilities based on the TAC library in this invention.

[0021] Figure 5 This is a data processing structure diagram of the multimodal early warning content adaptive adaptation engine of the present invention.

[0022] Figure 6 This is a schematic diagram of the protocol stack mapping and data packet structure for different terminal types according to the present invention.

[0023] Figure 7 This is a timing diagram for priority-based multi-channel concurrent distribution in this invention.

[0024] Figure 8 This is a closed-loop flowchart of the hierarchical degradation and blind spot feedback mechanism of the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] Combination Figures 1 to 8 The following is a detailed introduction to a multimodal fusion early warning information distribution method and system based on sudden meteorological disasters on highways, according to the present invention.

[0027] 1. Overall Implementation Process and System Architecture (corresponding to) Figure 1 , Figure 2 ) like Figure 1 As shown, the method provided by this invention mainly includes four core steps: S101 multi-source perception and audience selection; S102 terminal capability identification based on TAC library; S103 multi-modal content adaptive adaptation; S104 multi-channel concurrent distribution and closed-loop feedback.

[0028] The execution of this method depends on Figure 2 The system architecture shown is a physical interface that establishes communication links between air, ground, vehicles, and people. (1) Perception Layer: As the "eyes" of the system, it is deployed along the highway. Specifically, it includes: visibility sensors and micro-weather stations deployed every 2 kilometers to collect temperature, humidity and visibility data; high-definition AI cameras deployed every 500 meters, which have defogging algorithms and image recognition capabilities to capture fog patterns and traffic density; and 5G-A base stations that reuse the operator's existing network, using their ISAC (Integrated Communication and Sensing) capabilities to detect the outlines of objects in low visibility through millimeter-wave radar echo characteristics.

[0029] (2) Network Layer: Based on the operator's 5G SA standalone network architecture. Among them, the access network (RAN) is responsible for terminal connection; the MEC edge computing nodes are deployed in the aggregation room closest to the highway, responsible for real-time data cleaning and low-latency decision-making; the core network (5GC) provides user authentication, mobility management and signaling control.

[0030] (3) Platform Layer: Deploy the core "Multimodal Fusion Early Warning Center". This center contains three key subsystems: TAC terminal capability database (for querying device profiles), content adaptation engine (for format conversion), and unified message distribution gateway (integrating SMSC, MMSC, and V2X Server interfaces).

[0031] (4) Terminal Layer: It covers all types of terminals currently in operation, from OBU (on-board unit) that supports C-V2X protocol to 5G smartphones that support RCS, and traditional feature phones that only support 2G / 3G networks.

[0032] 2. Step S101: Multi-source perception and dynamic audience selection (corresponding to...) Figure 3 ) This step aims to address the questions of "when to send" and "to whom to send".

[0033] (1) Multi-source data weighted fusion determination algorithm like Figure 3 As shown, the system receives real-time access to three data sources: meteorological data. (Normalized visibility values), visual data (Fog obstruction rate identified by image recognition), radio frequency data D 3 (Millimeter-wave channel attenuation characteristics).

[0034] The system sets the warning trigger threshold. (For example, 50 meters). The weighted average method is used to calculate the comprehensive visibility index. : ; Among them, the weight vector The settings are dynamically adjusted based on sensor confidence levels. In this embodiment, considering that the optical camera is susceptible to interference in foggy scenes, the following settings are configured: .

[0035] When the calculation result When the system detects a "red-level fog disaster" on a road section (e.g., from K120+500 to K125+000 on the G3 expressway), it will immediately determine that such a section has been affected, trigger an early warning, and generate a unique disaster event ID.

[0036] (2) Location-based high-speed mobile audience targeting The system uses the determined center coordinates of the fog patch Using the fog diffusion model as the center, the radius of the electronic fence is set. (e.g., 5km).

[0037] Using the LMF (Location Management Function) component of the operator's core network, a list of all users currently within the electronic fence is retrieved in real time. To exclude nearby static users (such as residents of villages near highways), the system combines base station handover signaling and Doppler shift data to calculate the user's velocity vector. .

[0038] The filtering logic is as follows: ①Condition A: The user is in RRC_CONNECTED (connected state) or RRC_IDLE (idle state) but has a location update within the fenced cell.

[0039] ②Condition B: User movement speed .

[0040] ③Condition C: The angle between the direction of movement and the coordinates of the center of the fog patch (i.e., heading towards the fog area).

[0041] Users whose IMSI / IMEI meets all three of the above conditions are added to the "high-risk audience pool".

[0042] 3. Step S102: Intelligent identification of terminal capabilities based on the TAC library (corresponding to...) Figure 4 ) This step aims to address the prerequisite question of "what format to send".

[0043] like Figure 4 As shown, the system iterates through each user in the "high-risk audience pool" and extracts the first 8 digits of their IMEI (International Mobile Equipment Identity), namely TAC (Type Allocation Code).

[0044] The system queries the "Terminal Capability Database" pre-installed on the platform. This database is synchronized with the official GSMA database and has been expanded with operator-defined fields. The query process is as follows: (1) First-level judgment (vehicle networking capability): Check the Device_Type field. If the value is "Automotive", "T-Box", or the Marketing_Name contains keywords such as "Huawei HiCar" or "Qualcomm 9150", and the SIM card's APN subscription information is a vehicle-to-everything (V2X) APN (such as v2x.nbiot), then it is classified as a "Class I terminal (intelligent connected vehicle system)". This type of terminal has the ability to automatically parse machine commands and control the vehicle via the CAN bus.

[0045] (2) Second-level judgment (rich media capability): If it's not a vehicle-mounted system, continue checking the OS_Version and RCS_Support fields. If the terminal supports Android 10+ or ​​iOS 14+, and the operator has enabled VoLTE and 5G messaging services (UP 2.4 protocol), then it's classified as a "Category II Terminal (5G Smartphone)". This type of terminal has a high-definition large screen capable of displaying maps and videos.

[0046] (3) Third-level judgment (basic ability): If none of the above conditions are met (e.g., older Nokia phones, children's watches, or terminals that only support 2G networks), then it is classified as a "Category III terminal (basic function phone)". This type of terminal only guarantees the most basic SMS and voice communication.

[0047] 4. Step S103: Multimodal content adaptive adaptation (corresponding to...) Figure 5 , Figure 6 ) This step is the core of the invention, which uses an adaptation engine to transform a single warning intent into three heterogeneous protocol data.

[0048] (1) Standardized metadata input The early warning center first generates a standard meta data packet in JSON format: { "EventID": "FOG_20231027_001", Level: "RED" "Location": {"Lat": 31.86, "Lon": 117.29, "Radius": 3000}, "Action": "SpeedLimit", Value: 40, Text: "Dense fog 3 kilometers ahead, visibility less than 50 meters, please slow down immediately." "MediaURL": "[http: / / cdn.isp.com / warn / fog_demo.mp4](http: / / cdn.isp.com / warn / fog_demo.mp4)" } (2) Multimodal adaptation output like Figure 5 and Figure 6 As shown, the adaptation engine executes three conversion processes in parallel: ①Targeting a type of terminal (V2X signaling): Parse the Action and Value in the metadata and map them to RSI (Road Side Information) messages encoded in the ASN.1 (Abstract Syntax NotationOne) standard.

[0049] ○Critical data frame: Extract Value: 40 and convert it to hexadecimal control word 0x28.

[0050] ○ Protocol encapsulation: Encapsulated in the DSMP (Dedicated Short Message Protocol) layer of the PC5 interface.

[0051] Features: Extremely small data packets (<1KB), extremely low latency (<20ms), which can be directly read by the vehicle's ADAS system and trigger "pulse braking" or dashboard warnings.

[0052] ②Targeting Type II Terminals (MaaP Cardization): Based on the GSMA RCS Universal Profile standard, generate XML / JSON data in the Chatbot Card structure.

[0053] ○ Layout: The top features a red warning banner, the middle section displays dynamic GIS map tiles generated based on the location, and the bottom displays a preview image of the live video corresponding to the MediaURL.

[0054] ○ Interaction: Embedded "One-Click Rescue" button (tel:12122) and "View Escape Route" button (DeepLink jumps to the map app).

[0055] Features: Large amount of information, intuitive and visual, convenient for the driver and co-driver to view in advance.

[0056] ③ Targeting three types of terminals (strong reminders): A "minimalist + strong interference" strategy is adopted.

[0057] ○USSD / Flash Message: Extract the Text field and encapsulate it as a Class 0 SMS. Unlike ordinary Class 1 SMS messages stored in the inbox, Class 0 message signaling, upon reaching the phone's baseband chip, will directly trigger a pop-up window on the screen and illuminate the backlight, allowing the user to read the message without unlocking the phone.

[0058] ○TTS Voice: Using iFlytek or a similar TTS engine, the text is converted into an 8kHz / 16bit PCM voice stream for subsequent outbound calls.

[0059] 5. Step S104: Multi-channel concurrent distribution and closed-loop blind patching (corresponding to...) Figure 7 , Figure 8 ) This step uses a sophisticated strategy to ensure that information is "delivered".

[0060] (1) Priority queues and concurrent dispatch The system establishes three sending queues, and sets the highest priority (Priority=1) for disaster events that trigger early warnings.

[0061] like Figure 7 As shown, at time T0, the system simultaneously calls three gateway interfaces: ①V2X Gateway: Sends RSI signaling to a type of terminal via roadside RSU (Roadside Unit) broadcast and Uu interface unicast.

[0062] ②5G Messaging Center (MaaP): Distributes rich media cards to Class II terminals via the IMS network.

[0063] ③ Signaling gateway: Sends Flash SMS messages to three types of terminals.

[0064] (2) Status monitoring and “degradation and blind spot filling” mechanism like Figure 8 As shown, to address information loss caused by network jitter or terminal malfunctions, the system employs robust blind spot compensation logic: ① Anchor point 1 (Cross-device blind spot filling from vehicle infotainment system to mobile phone): The system starts a 200ms timer after sending V2X signaling. If no application layer acknowledgment (ACK) is received from the vehicle's infotainment system within 200ms, the system is considered offline or the OBU is faulty.

[0065] Compensation action: Immediately retrieve the vehicle owner's mobile phone number bound to the vehicle's infotainment system, add the number to the "Class II Terminal" sending queue, and send a 5G message to ensure that the vehicle owner can receive the notification.

[0066] ② Anchor Point 2 (Cross-domain gap filling from data to speech): The system starts a 5-second timer after sending a 5G message. If a status report is received indicating "Failed" or "Expired," it usually means that the user is in a weak network environment (the data channel is blocked, but the voice channel may be normal).

[0067] Compensation Action: Immediately trigger the IVR (Interactive Voice Response) interface of the core network to initiate an automatic outbound call based on VoLTE. After the call is connected, directly play the TTS voice script generated in step S103. This method has the strongest penetration and can maximize the driver's attention.

[0068] 6. Application Scenarios and Examples Scene background: The implementation process of this invention is illustrated by taking an example of a sudden localized fog on a certain section of a highway at 7:00 AM on a certain day in November of a certain year, in which visibility dropped sharply to 30 meters.

[0069] Implementation process: (1) Sensing link: The visibility meter deployed on the roadside at K110 reported a value of 28m. At the same time, the 5G base station sensed the abnormal attenuation of radio frequency signal in the area and determined that it was a disaster warning event.

[0070] (2) Selection process: The platform selected 50 users whose vehicles were about to enter the area.

[0071] (3) Identification and adaptation process: The system identifies a user's vehicle as having an intelligent connected system and classifies it as a first-class terminal; it identifies another user as using a mainstream brand smartphone and classifies it as a second-class terminal; and it identifies yet another user as using a traditional feature phone and classifies it as a third-class terminal.

[0072] (4) Distribution and execution phase: For the first type of end user, a red pop-up window displaying "Speed ​​Limit Ahead Due to Fog" appears directly on the vehicle's central control screen, and the cruise speed is automatically reduced. For the second type of end user, the mobile phone receives a 5G rich media card displaying a map of the fog area ahead. If the acknowledgment is not received in time due to signal fluctuations, the system automatically triggers a voice call, and the user receives an alarm call and slows down. For the third type of end user, the mobile phone screen lights up directly, displaying a plain text warning window accompanied by vibration.

[0073] As can be seen from the above embodiments, the present invention can provide differentiated and highly reliable early warning services for terminals with different capability levels and complex network environments, effectively making up for the shortcomings of traditional methods.

[0074] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for distributing multimodal fusion early warning information under sudden meteorological disasters on highways, characterized in that, include: (1) Construct a multi-source perception and audience selection mechanism for sudden meteorological disasters By integrating meteorological bureau micro-monitoring data, highway roadside sensor data, and operator base station integrated sensing data, meteorological disaster information is obtained, and meteorological disasters that trigger early warnings are identified. Based on operator base station positioning technology, a dynamic electronic fence is established to select all mobile communication terminal users who are about to enter or are already in the disaster-affected area in real time, and a target audience list is generated. (2) Intelligent identification and classification of terminal capabilities based on TAC library By using the device identifier register or big data analysis platform of the operator's core network, the first 8 digits of the TAC of each IMEI in the audience target list are parsed, and the terminal capability database is queried to determine the type of the audience terminal; (3) Establish a multimodal early warning content adaptive adaptation engine Standardized meteorological warning metadata is pre-set, and heterogeneous protocol data adapted to the type of the target terminal is generated based on the type of the target terminal. (4) Perform priority-based multi-channel concurrent and blind-spot distribution. Develop a priority-based multi-channel concurrent retransmission strategy, monitor the transmission status of each channel in real time, and retransmit failed targets across channels to ensure that warning information is "zero omission".

2. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 1, characterized in that, The process of determining the meteorological disaster that triggers the warning is as follows: The weighted average method is used to calculate the comprehensive visibility index. : ； in, This is the normalized visibility value; The fog obstruction rate identified in the image; D 3 represents the attenuation characteristic value of the millimeter-wave channel; , and They are respectively , and D The weighting factor is 3; Set warning trigger threshold When the comprehensive visibility index Less than the warning trigger threshold At that time, an early warning is triggered.

3. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 1, characterized in that, In step (1), the mobile communication terminal users are selected as follows: Using the location management function component of the operator's core network, a list of all users currently within the electronic fence is retrieved in real time. To eliminate nearby static users, the velocity vector of each user is calculated by combining base station handover signaling and Doppler frequency shift data. ; The selection logic is as follows: ①Condition A: The user is in a connected or idle state but has a location update within the fenced cell; ②Condition B: Set a movement speed threshold; the user's movement speed must be greater than the movement speed threshold. ③Condition C: The angle between the user's movement direction and the coordinates of the center of the fog patch. ; Users who meet all three of the above conditions are selected for IMSI / IMEI.

4. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 1, characterized in that, The types of audience terminals include: The first type of terminal is the "intelligent connected vehicle unit", which has V2X communication and vehicle control capabilities; the second type of terminal is the "5G intelligent handheld terminal", which supports 5G messaging and high-definition video playback; the third type of terminal is the "basic function terminal", which only supports voice and ordinary text messages.

5. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 4, characterized in that, For Type 1 terminals, the engine encapsulates metadata into V2X standard control signaling, including speed limit thresholds and braking commands; for Type 2 terminals, the engine automatically generates 5G rich media cards containing dynamic maps, on-site GIF animations, and text descriptions; for Type 3 terminals, the engine extracts core text information to generate flash message text that is forced to pop up or generates emergency call scripts through TTS.

6. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 5, characterized in that, The priority-based multi-channel concurrent re-processing strategy and blind spot distribution in step (4) are as follows: Establish three sending queues and simultaneously call three gateway interfaces: ①V2X Gateway: Sends RSI signaling to a type of terminal via roadside RSU broadcast and Uu interface unicast; ②5G Messaging Center: Distributes rich media cards to Class II terminals via the IMS network; ③ Signaling gateway: Sends Flash SMS messages to three types of terminals; V2X signaling is prioritized for Class I terminals, while 5G messages are sent to Class II terminals bound to the vehicle. 5G rich media cards are prioritized for Class 1 terminals, while Class 2 terminals that have not received a confirmation receipt are automatically downgraded and triggered to make a strong reminder via intelligent voice call (IVR). For the three types of terminals, the flash message domination method is directly adopted.

7. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 5, characterized in that, The blind spot filling logic is as follows: ① Anchor point 1, cross-platform blind spot filling between vehicle infotainment system and mobile phone: After sending V2X signaling, a timer is started. If no application layer confirmation is received from the vehicle's infotainment system within the specified time, the system is determined to be offline or the OBU is faulty. The vehicle owner's mobile phone number bound to the infotainment system is immediately retrieved, the number is added to the "Class II Terminal" sending queue, and a 5G message is sent. ② Anchor point 2, cross-domain gap filling from data to speech: After sending a 5G message, the system starts a timer. If a status report is received indicating that the transmission failed or timed out, it usually means that the user is in a weak network environment. The interactive voice response interface of the core network is immediately triggered to initiate an automatic outbound call based on VoLTE. After the call is connected, the generated TTS voice script is played directly.

8. The multimodal fusion early warning information distribution method based on sudden meteorological disasters on highways according to claim 1, characterized in that, The meteorological early warning metadata includes disaster type, location coordinates, recommended vehicle speed, evacuation route, and on-site video stream.

9. A system for implementing the method according to any one of claims 1-8, characterized in that, include: The perception layer includes several visibility meters and micro-weather stations deployed at intervals along the highway to collect temperature, humidity, and visibility data; high-definition AI cameras deployed at intervals along the highway, equipped with defogging algorithms and image recognition capabilities, to capture fog patterns and traffic density; and reuse of existing 5G-A base stations from operators' networks, utilizing their integrated communication and perception capabilities to detect object outlines in low visibility conditions through millimeter-wave radar echo characteristics. Network Layer: Based on the operator's 5G SA standalone network architecture, the access network is responsible for terminal connection; MEC edge computing nodes are deployed in the aggregation room closest to the highway, responsible for real-time data cleaning and low-latency decision-making; the core network provides user authentication, mobility management and signaling control. Platform layer: Deploy the core "Multimodal Fusion Early Warning Center", which includes three key subsystems: TAC terminal capability database, used to query device profiles; content adaptation engine, used for format conversion; and unified message distribution gateway, which integrates SMSC, MMSC, and V2X Server interfaces. Terminal layer: It covers all types of terminals currently in operation, from vehicle units that support the C-V2X protocol to 5G smartphones that support RCS, and traditional feature phones that only support 2G / 3G networks.

Images