A vehicle data communication method and system based on cloud collaboration

Abstract

This invention discloses a cloud-based collaborative vehicle data communication method and system, comprising: Step 1: collecting vehicle communication-related data and data generation timestamps to generate vehicle data value tags; Step 2: constructing a vehicle communication state matrix based on communication link data; Step 3: generating link mutation risk values ​​using an improved SINDy model; Step 4: generating regional communication intent codes and distributing them to roadside edge nodes; Step 5: generating data window matching results; Step 6: performing hierarchical slicing processing on vehicle service data and determining the sending action; Step 7: performing vehicle service data transmission processing and updating the regional communication intent codes. This invention achieves dynamic collaborative communication control and adaptive scheduling of link risks for vehicle data.

Description

Technical Field

[0001] This invention relates to the fields of vehicle networking communication and machine learning technology, and in particular to a vehicle data communication method and system based on cloud collaboration. Background Technology

[0002] With the rapid development of vehicle-to-everything (V2X) technology, edge computing technology, and cloud-based collaborative communication technology, the scale of data generated during vehicle operation continues to grow. The frequency of data interaction between vehicle terminals, roadside edge nodes, and cloud servers is constantly increasing. The real-time transmission capability of vehicle business data is gradually becoming a crucial factor affecting intelligent driving, collaborative perception, and vehicle-road cooperative control. For vehicle data communication scenarios, existing technologies mainly employ fixed-priority transmission, static queue scheduling, or dynamic bandwidth allocation based on network state parameters to perform vehicle data transmission control. However, these methods generally suffer from the following problems in complex and dynamic road environments: Most existing vehicle data communication methods perform link scheduling based on instantaneous network conditions, lacking the ability to continuously model the dynamic evolution of vehicle communication links. This makes it difficult to identify sudden link changes under conditions such as high-speed vehicle movement, edge node load variations, and road event disturbances, leading to problems such as transmission delay fluctuations and decreased link stability during vehicle data transmission. Existing communication scheduling methods typically perform single-dimensional scheduling based on data priority or network load, lacking a joint modeling mechanism for the correlation between link risk status, cloud receiving window, and regional communication resource status, making it difficult to achieve dynamic matching of data windows in complex communication scenarios. Furthermore, most existing vehicle data transmission methods adopt a complete data upload approach, lacking a layered slicing mechanism based on the structural characteristics of vehicle business data. When facing link fluctuations and resource-constrained scenarios, this can easily lead to critical state data transmission failures or excessive edge resource consumption. At the same time, current communication control methods lack a regional communication feedback update mechanism based on historical reception validity, making it difficult to form a dynamic collaborative communication closed loop between vehicles, edge nodes, and the cloud.

[0003] Therefore, how to provide a cloud-based collaborative vehicle data communication method and system is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] One objective of this invention is to propose a cloud-based collaborative vehicle data communication method and system. This invention achieves dynamic evolution modeling and link mutation risk identification of vehicle communication links by constructing a phase inheritance residual triggering switching SINDy model, regional communication intent codes, and a vehicle service data hierarchical slicing mechanism. Combined with an action-sensitive sparse competition structure, a data window matching mechanism, and a historical reception validity update mechanism, it realizes adaptive collaborative communication control between vehicle terminals, roadside edge nodes, and the cloud, thereby improving the stability of vehicle data transmission and the ability to schedule communication resources.

[0005] A cloud-based collaborative vehicle data communication method according to an embodiment of the present invention includes the following steps: Step 1: Collect vehicle communication-related data and data generation timestamps to generate vehicle data value tags; Step 2: Construct a vehicle communication state matrix based on the communication link data in the vehicle communication association data; Step 3: Construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate link mutation risk values ​​based on the sparse coefficient matrices corresponding to adjacent link stages. Step 4: Generate a regional communication intent code based on the link mutation risk value and vehicle communication association data, and send it to the roadside edge node; Step 5: Generate a data window matching result based on the link mutation risk value, vehicle data value tag, predicted transmission completion time, and cloud receiving window in the regional communication intent code; Step 6: Based on the vehicle data value tag, perform hierarchical slicing processing on the vehicle business data in the vehicle communication association data to generate core slices and supplementary slices, and determine the sending action corresponding to the vehicle business data by combining the link mutation risk value, data window matching result and regional communication intent code. Step 7: Perform vehicle service data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

[0006] Optionally, step one includes: It collects vehicle business data, vehicle status data, communication link data, regional vehicle density data, edge node load data, cloud queue data, and road event data during vehicle operation, and generates timestamps based on the collection time. The vehicle business data is divided into three categories: safety event data, status synchronization data, and background awareness data. Corresponding data type identifiers are then established based on these categories. Perform state change detection processing on vehicle state data, calculate the changes in speed, acceleration, and steering angle between adjacent acquisition times, and generate a vehicle dynamic state identifier based on the state change. Perform link timing processing on the communication link data, and arrange the signal-to-noise ratio, round-trip delay, number of retransmissions and packet loss rate in time series according to the data generation timestamp to form a communication link state sequence; Spatial clustering is performed on regional vehicle density data to generate regional vehicle density distribution results based on vehicle location coordinates and distances between vehicles. Interval statistical processing is also performed on edge node load data and cloud queue data according to time series to form an edge resource status sequence. The road event data is processed for event association. Based on the location of the road event, the duration of the event and the road segment associated with the event, a road event impact identifier is generated, and time sequence alignment is performed according to the data generation timestamp and the communication link status sequence. Based on data type identifiers, vehicle dynamic status identifiers, communication link status sequences, regional vehicle density distribution results, edge resource status sequences, and road event impact identifiers, state association coding processing is performed to generate vehicle data value tags.

[0007] Optionally, step two includes: Extract communication link data from vehicle communication association data, and perform time-series sorting processing on the communication link data according to the data generation timestamp; Based on the time-series sorted communication link data, the signal-to-noise ratio, round-trip delay, number of retransmissions, packet loss rate, and edge node buffer queue length are extracted, and data alignment processing is performed according to the collection time. The signal-to-noise ratio (SNR) is divided into fluctuation intervals, the SNR change between adjacent acquisition times is calculated, and an SNR change sequence is generated. Perform time series correlation permutation processing on round-trip delay, retransmission count, and packet loss rate to generate link delay change sequence, retransmission change sequence, and packet loss change sequence; The relative displacement between vehicles is calculated based on the vehicle's position coordinates and the position coordinates of adjacent vehicles, and a relative speed sequence of adjacent vehicles is generated based on the relative displacement between vehicles. Perform time-series permutation processing on the length of the edge node cache queue to generate a change sequence of the edge cache queue; Based on the signal-to-noise ratio change sequence, link delay change sequence, retransmission change sequence, packet loss change sequence, relative speed sequence of neighboring vehicles, and edge buffer queue change sequence, a vehicle communication state matrix is ​​generated by performing vector combination processing according to a unified time index.

[0008] Optionally, step three includes: A candidate function library is constructed based on the vehicle communication state matrix. The candidate function library includes signal-to-noise ratio variation items, link delay variation items, retransmission variation items, packet loss variation items, relative speed variation items of neighboring vehicles, edge buffer queue variation items, as well as product coupling items and timing differential coupling items between various state variables. The candidate function library is input into the improved SINDy model of stage inheritance residual triggering switching. The improved SINDy model of stage inheritance residual triggering switching includes a stage sparse identification structure, a residual triggering switching structure, a stage inheritance sparse structure, and an action-sensitive sparse competition structure. The action-sensitive sparse competition structure is used to adjust the sparse competition relationship between candidate function items based on the historical communication action type. The candidate function library is partitioned into candidate groups using the aforementioned action-sensitive sparse competition structure, including: Based on the historical communication action types in the vehicle communication association data, the candidate function items are divided into upload association candidate group, relay association candidate group, cache association candidate group and slice association candidate group; Based on the candidate group corresponding to the current historical communication action type, the sparse competition order between candidate function items is adjusted, and sparse coefficient compression processing is performed on the adjusted candidate function items to generate a set of action-related candidate functions. The sparse dynamics identification process is performed on the action association candidate function set using the stage sparse identification structure. Based on the candidate function terms, the sparse coefficients are solved for the dynamic evolution relationship of the state variables in the vehicle communication state matrix to generate the sparse coefficient matrix corresponding to the current link stage. The residual-triggered switching structure is used to perform dynamic residual detection processing on the sparse coefficient matrix corresponding to the current link stage, including: The state prediction result is calculated based on the sparse coefficient matrix corresponding to the current link stage. Calculate the state residual sequence based on the state prediction results and the actual state values ​​in the vehicle communication state matrix; Perform time window accumulation processing on the state residual sequence to generate a residual change sequence; Based on the growth segment of the residual change sequence in the continuous time interval, perform link stage switching detection and generate a new link stage identifier; The stage inheritance sparse structure is used to perform stage inheritance constraint processing on the sparse coefficient matrix corresponding to the new link stage, including: Extract the non-zero sparse coefficients corresponding to the current link stage; Based on the non-zero sparse coefficient terms, perform sparse constraint update processing on the candidate function terms in the new link stage; Based on the updated candidate function terms, the sparse coefficient solution process is re-executed to generate the sparse coefficient matrix corresponding to the new link stage. Sparse difference calculation is performed on the sparse coefficient matrix corresponding to adjacent link stages to generate a sparse change matrix. Based on the non-zero change terms in the sparse change matrix, the state offset between link stages is calculated to generate a link mutation risk value.

[0009] Optionally, step four includes: Extract regional vehicle density data, edge node load data, cloud queue data, and road event data from vehicle communication association data, and perform time-series association and sorting processing according to the data generation timestamp; Spatial aggregation and correlation processing is performed on regional vehicle density data to generate a vehicle density change sequence based on the distance change relationship between vehicle location coordinates. Perform resource time-series correlation processing on edge node load data and cloud queue data, and generate a resource pressure change sequence based on the changes in edge node load and cloud queue length within a continuous time interval; Perform event impact correlation processing on road event data to generate road event impact sequences based on the location, duration, and associated road segments of the road events. Based on the link mutation risk value, vehicle density change sequence, resource pressure change sequence, and road event impact sequence, multi-dimensional state coding processing is performed to generate a regional communication state vector. Perform state interval partitioning on the regional communication state vector to generate communication scheduling interval identifiers for the corresponding state intervals; Based on the communication scheduling interval identifier, link mutation risk value, and resource pressure change sequence, interval coding processing is performed to generate a regional communication intent code including the scheduling interval identifier, risk interval identifier, and resource status interval identifier, and the regional communication intent code is sent to the corresponding roadside edge node.

[0010] Optionally, step five includes: Extract the scheduling interval identifier, risk interval identifier, and resource status interval identifier from the regional communication intent code, and extract the start and end times of the corresponding cloud receiving window; The link transmission delay change is calculated based on the link mutation risk value, resource status interval identifier, and link delay change sequence in the vehicle communication status matrix, and a predicted transmission duration is generated based on the link transmission delay change. The predicted transmission completion time is calculated based on the data generation timestamp and the predicted transmission duration. Perform value range segmentation processing on the vehicle data value tags to generate data value range identifiers for the corresponding vehicle business data; Perform time offset correlation processing based on the time offset between the predicted transmission completion time and the cloud receiving window to generate a window time offset sequence; Window state encoding is performed based on the window time series offset sequence, data value interval identifier, and risk interval identifier to generate a data window associated state vector. Perform window matching interval partitioning processing on the data window associated state vector to generate data window matching results and corresponding matching priority identifiers.

[0011] Optionally, step six includes: Extract vehicle business data from vehicle communication associated data, and perform value range classification processing on vehicle business data based on vehicle data value tags to generate high-value data range, medium-value data range and low-value data range; Perform field association splitting processing on vehicle business data, and divide key status fields and extended status fields based on the temporal association and transmission dependency between data fields; Core slices are generated based on key status fields, supplementary slices are generated based on extended status fields, and slice association indexes are established between core slices and supplementary slices according to the data generation timestamp. Extract the matching priority identifier, scheduling interval identifier, and risk interval identifier from the data window matching results, and perform status association sorting processing according to the unified time index; Based on the link mutation risk value, matching priority identifier, scheduling interval identifier, and risk interval identifier, action status encoding processing is performed to generate a sending action status vector. Perform action interval partitioning processing on the sending action state vector to generate the sending action type corresponding to the vehicle service data; Based on the sending action type, the core slice and supplementary slice are processed to perform action association mapping, and the corresponding vehicle service data sending action is generated.

[0012] Optionally, step seven includes: Extract the sending actions corresponding to vehicle service data, and perform transmission order sorting processing on the core slice and supplementary slice based on the sending action type; Establish a corresponding data transmission queue based on the sending action type, and perform queue writing processing on the core slice and supplementary slice according to the data generation timestamp; Perform transmission status recording processing on the core slices and supplementary slices in the data transmission queue, recording the upload completion time, retransmission count, queue waiting time, and cloud reception status of the corresponding slices; A transmission delay sequence is generated based on the time difference between the upload completion time and the data generation timestamp, and a transmission stability sequence is generated based on the number of retransmissions and the cloud reception status. Based on the transmission delay sequence, transmission stability sequence, and cloud reception status, perform reception validity encoding processing to generate historical reception validity data; Perform time-interval statistical processing on historical reception validity data to generate a historical validity change sequence; Based on the historical validity change sequence, the resource status interval identifier and risk interval identifier in the regional communication intent code, the interval status update process is performed to generate the updated regional communication intent code and send it to the corresponding roadside edge node.

[0013] A cloud-based collaborative vehicle data communication system according to an embodiment of the present invention includes: The data acquisition and value labeling module is used to collect vehicle communication-related data and data generation timestamps, and generate vehicle data value tags. The state matrix construction module is used to construct the vehicle communication state matrix based on the communication link data in the vehicle communication associated data. The link risk identification module is used to construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate a link mutation risk value based on the sparse coefficient matrix corresponding to the current link stage. The regional intent code generation module is used to generate regional communication intent codes based on link mutation risk values ​​and vehicle communication association data, and then distribute them to roadside edge nodes. The window matching analysis module is used to generate data window matching results based on the link mutation risk value, vehicle data value label, predicted transmission completion time, and cloud receiving window in the regional communication intent code. The slice scheduling decision module is used to perform hierarchical slicing processing on vehicle business data in vehicle communication associated data based on vehicle data value tags, generate core slices and supplementary slices, and determine the sending action corresponding to vehicle business data by combining link mutation risk value, data window matching result and regional communication intent code. The transmission feedback update module is used to perform vehicle business data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

[0014] The beneficial effects of this invention are: This invention addresses the challenges of continuously modeling dynamic changes in link states, identifying communication link mutations in advance, and lacking dynamic scheduling capabilities in complex vehicle-to-everything (V2X) scenarios by constructing a collaborative communication mechanism among vehicle communication association data, regional communication intent codes, and data window matching results. It combines a stage-inherited residual-triggered switching SINDy model with a joint design of an action-sensitive sparse competition structure. The invention proposes a communication dynamics modeling strategy based on link stage switching, sparse competition relationship adjustment, and stage inheritance constraint updates. This significantly improves the ability to identify vehicle communication link mutation risks and model dynamic evolution of link states. Furthermore, an action-sensitive sparse competition structure is introduced in the link risk identification stage. By adjusting the sparse competition order among candidate function terms based on historical communication action types, the link change characteristics corresponding to different communication behaviors are realized. The system employs adaptive sparse identification; in the regional collaborative communication phase, it constructs a regional communication intent code, and through joint encoding of link mutation risk values, resource pressure change sequences, and road event impact sequences, it achieves dynamic expression of regional communication status and edge collaborative scheduling; in the data transmission phase, it constructs a vehicle business data hierarchical slicing mechanism, and through the associated scheduling between core slices and supplementary slices, it achieves priority transmission of key status data and dynamic allocation of communication resources; in the feedback update phase, it constructs a historical reception validity update mechanism, and combines transmission delay sequences, transmission stability sequences, and cloud reception status to perform dynamic updates of the regional communication intent code, forming a closed-loop collaborative communication control between vehicle terminals, roadside edge nodes, and cloud servers, realizing dynamic risk perception, adaptive window matching, and highly stable transmission of vehicle business data. Attached Figure Description

[0015] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of a cloud-based collaborative vehicle data communication method proposed in this invention; Figure 2 This is a schematic diagram of the structure of a cloud-based collaborative vehicle data communication system proposed in this invention. Detailed Implementation

[0016] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0017] refer to Figure 1 A cloud-based collaborative vehicle data communication method includes the following steps: Step 1: Collect vehicle communication-related data and data generation timestamps to generate vehicle data value tags; Step 2: Construct a vehicle communication state matrix based on the communication link data in the vehicle communication association data; Step 3: Construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate link mutation risk values ​​based on the sparse coefficient matrices corresponding to adjacent link stages. Step 4: Generate regional communication intent codes based on link mutation risk values ​​and vehicle communication association data, and send them to roadside edge nodes; Step 5: Generate data window matching results based on link mutation risk value, vehicle data value label, predicted transmission completion time, and cloud receiving window in regional communication intent code; Step 6: Perform hierarchical slicing processing on vehicle business data in vehicle communication association data based on vehicle data value tags to generate core slices and supplementary slices, and determine the sending action corresponding to vehicle business data by combining link mutation risk value, data window matching result and regional communication intent code. Step 7: Perform vehicle service data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

[0018] In this embodiment, step one includes: It collects vehicle business data, vehicle status data, communication link data, regional vehicle density data, edge node load data, cloud queue data, and road event data during vehicle operation, and generates timestamps based on the collection time. The vehicle business data is divided into three categories: safety event data, status synchronization data, and background awareness data. Corresponding data type identifiers are then established based on these categories. Perform state change detection processing on vehicle state data, calculate the changes in speed, acceleration, and steering angle between adjacent acquisition times, and generate a vehicle dynamic state identifier based on the state change. Perform link timing processing on the communication link data, and arrange the signal-to-noise ratio, round-trip delay, number of retransmissions and packet loss rate in time series according to the data generation timestamp to form a communication link state sequence; Spatial clustering is performed on regional vehicle density data to generate regional vehicle density distribution results based on vehicle location coordinates and distances between vehicles. Interval statistical processing is also performed on edge node load data and cloud queue data according to time series to form an edge resource status sequence. The road event data is processed for event association. Based on the location of the road event, the duration of the event and the road segment associated with the event, a road event impact identifier is generated, and time sequence alignment is performed according to the data generation timestamp and the communication link status sequence. Based on data type identifiers, vehicle dynamic status identifiers, communication link status sequences, regional vehicle density distribution results, edge resource status sequences, and road event impact identifiers, state association coding processing is performed to generate vehicle data value tags.

[0019] In this implementation, vehicle business data includes autonomous driving control data, vehicle perception data, and vehicle cooperative interaction data; vehicle status data includes vehicle speed, vehicle acceleration, steering angle, and braking status; and communication link data includes data transmission status information between the vehicle and roadside edge nodes. The status association coding process involves performing temporal association and arrangement on data type identifiers, vehicle dynamic status identifiers, communication link status sequences, regional vehicle density distribution results, edge resource status sequences, and road event impact identifiers, and generating a multi-dimensional status association vector according to a unified time base. The regional vehicle density distribution results are divided into multiple vehicle clustering areas based on the distance changes between vehicle position coordinates, and the edge resource status sequences generate corresponding resource change sequences based on the edge node load changes and cloud queue changes within a continuous time interval.

[0020] In this embodiment, step two includes: Extract communication link data from vehicle communication association data, and perform time-series sorting processing on the communication link data according to the data generation timestamp; Based on the time-series sorted communication link data, the signal-to-noise ratio, round-trip delay, number of retransmissions, packet loss rate, and edge node buffer queue length are extracted, and data alignment processing is performed according to the collection time. The signal-to-noise ratio (SNR) is divided into fluctuation intervals, the SNR change between adjacent acquisition times is calculated, and an SNR change sequence is generated. Perform time series correlation permutation processing on round-trip delay, retransmission count, and packet loss rate to generate link delay change sequence, retransmission change sequence, and packet loss change sequence; The relative displacement between vehicles is calculated based on the vehicle's position coordinates and the position coordinates of adjacent vehicles, and the relative speed sequence of adjacent vehicles is generated based on the relative displacement between vehicles. Perform time-series permutation processing on the length of the edge node cache queue to generate a change sequence of the edge cache queue; Based on the signal-to-noise ratio change sequence, link delay change sequence, retransmission change sequence, packet loss change sequence, relative speed sequence of neighboring vehicles, and edge buffer queue change sequence, a vehicle communication state matrix is ​​generated by performing vector combination processing according to a unified time index.

[0021] In this implementation, the communication link data originates from the data interaction records between the vehicle terminal and the roadside edge nodes. The signal-to-noise ratio, round-trip delay, retransmission count, and packet loss rate are written into the corresponding time index according to a unified collection period. The relative displacement between vehicles is calculated based on the difference in vehicle position coordinates and the difference in neighboring vehicle position coordinates between consecutive collection times. The relative speed sequence of neighboring vehicles is generated according to the displacement change rate between adjacent time indices. The edge buffer queue change sequence is generated based on the change in buffer queue length within a continuous time interval. Vector combination processing includes performing dimensional mapping and arrangement of the signal-to-noise ratio change sequence, link delay change sequence, retransmission change sequence, packet loss change sequence, relative speed sequence of neighboring vehicles, and edge buffer queue change sequence according to a unified time index, and writing them into the corresponding matrix rows of the vehicle communication state matrix in chronological order.

[0022] In this embodiment, step three includes: A candidate function library is constructed based on the vehicle communication state matrix. The candidate function library includes signal-to-noise ratio variation, link delay variation, retransmission variation, packet loss variation, relative speed variation of neighboring vehicles, edge buffer queue variation, as well as product coupling terms and timing differential coupling terms between various state variables. The candidate function library is input into the improved SINDy model of stage inheritance residual triggering switching. The improved SINDy model of stage inheritance residual triggering switching includes a stage sparse identification structure, a residual triggering switching structure, a stage inheritance sparse structure, and an action-sensitive sparse competition structure. The action-sensitive sparse competition structure is used to adjust the sparse competition relationship between candidate function items based on the historical communication action type. The candidate function library is partitioned into candidate groups using an action-sensitive sparse competitive structure, including: Based on the historical communication action types in the vehicle communication association data, the candidate function items are divided into upload association candidate group, relay association candidate group, cache association candidate group and slice association candidate group; Based on the candidate group corresponding to the current historical communication action type, the sparse competition order between candidate function items is adjusted, and sparse coefficient compression processing is performed on the adjusted candidate function items to generate a set of action-related candidate functions. The sparse dynamics identification process is performed on the action association candidate function set using the stage sparse identification structure. Based on the candidate function terms, the sparse coefficients are solved for the dynamic evolution relationship of the state variables in the vehicle communication state matrix to generate the sparse coefficient matrix corresponding to the current link stage. The residual-triggered switching structure is used to perform dynamic residual detection processing on the sparse coefficient matrix corresponding to the current link stage, including: The state prediction result is calculated based on the sparse coefficient matrix corresponding to the current link stage. Calculate the state residual sequence based on the state prediction results and the actual state values ​​in the vehicle communication state matrix; Perform time window accumulation processing on the state residual sequence to generate a residual change sequence; Based on the growth segment of the residual change sequence in the continuous time interval, perform link stage switching detection and generate a new link stage identifier; The stage inheritance sparse structure is used to perform stage inheritance constraint processing on the sparse coefficient matrix corresponding to the new link stage, including: Extract the non-zero sparse coefficients corresponding to the current link stage; Sparse constraint update processing is performed on candidate function terms in the new link stage based on non-zero sparse coefficient terms; Based on the updated candidate function terms, the sparse coefficient solution process is re-executed to generate the sparse coefficient matrix corresponding to the new link stage. Sparse difference calculation is performed on the sparse coefficient matrix corresponding to adjacent link stages to generate a sparse change matrix. Based on the non-zero change terms in the sparse change matrix, the state offset between link stages is calculated to generate the link mutation risk value.

[0023] In this implementation, the improved SINDy model is constructed based on the sparse identification nonlinear dynamic model. It follows the processing method of the SINDy model to establish a candidate function library for state variables and perform sparse coefficient solution. The sparse coefficient matrix is ​​used to characterize the dynamic evolution relationship between vehicle communication state variables, and the main dynamic characteristics in the vehicle communication link are described based on non-zero sparse terms.

[0024] The improved SINDy model introduces a stage inheritance residual-triggered switching structure and an action-sensitive sparse competition structure on the basis of the SINDy model. The residual-triggered switching structure performs link stage switching detection based on the changing trend of the state residual sequence in a continuous time interval. The stage inheritance sparse structure performs sparse constraint update processing on the new link stage based on the non-zero sparse coefficients in the current link stage. The action-sensitive sparse competition structure adjusts the sparse competition order between candidate function terms based on the historical communication action types.

[0025] The improved SINDy model uses a link phase switching mechanism to distinguish between steady-state and abrupt-change phases in vehicle communication links. It utilizes a phase inheritance sparse structure to maintain the dynamic continuity between adjacent link phases and an action-sensitive sparse competition structure to enhance the sparse identification capability of link change features corresponding to different communication actions, thereby improving the accuracy of identifying abrupt-change risks in vehicle communication links.

[0026] In this embodiment, step four includes: Extract regional vehicle density data, edge node load data, cloud queue data, and road event data from vehicle communication association data, and perform time-series association and sorting processing according to the data generation timestamp; Spatial aggregation and correlation processing is performed on regional vehicle density data to generate a vehicle density change sequence based on the distance change relationship between vehicle location coordinates. Perform resource time-series correlation processing on edge node load data and cloud queue data, and generate a resource pressure change sequence based on the changes in edge node load and cloud queue length within a continuous time interval; Perform event impact correlation processing on road event data to generate road event impact sequences based on the location, duration, and associated road segments of the road events. Multidimensional state coding is performed based on link mutation risk value, vehicle density change sequence, resource pressure change sequence and road event impact sequence to generate regional communication state vector; Perform state interval partitioning on the regional communication state vector to generate communication scheduling interval identifiers for the corresponding state intervals; Based on the communication scheduling interval identifier, link mutation risk value, and resource pressure change sequence, interval coding processing is performed to generate a regional communication intent code that includes the scheduling interval identifier, risk interval identifier, and resource status interval identifier, and the regional communication intent code is sent to the corresponding roadside edge node.

[0027] In this implementation, the regional communication state vector is generated by dimensionally arranging the vehicle density change sequence, resource pressure change sequence, road event impact sequence, and link mutation risk value according to a unified time index; the communication scheduling interval identifier is divided into multiple communication state intervals according to the state change amplitude in the regional communication state vector, and different communication state intervals correspond to different data transmission priorities; the risk interval identifier is generated based on the change trend of the link mutation risk value in a continuous time interval, and the resource state interval identifier is generated based on the combination relationship between the edge node load change and the cloud queue length change; the regional communication intent code is written into the scheduling interval identifier, risk interval identifier, and resource state interval identifier using a multi-field combination encoding method, and is processed accordingly according to the coverage of the regional roadside edge nodes.

[0028] In this embodiment, step five includes: Extract the scheduling interval identifier, risk interval identifier, and resource status interval identifier from the regional communication intent code, and extract the start and end times of the corresponding cloud receiving window; The link transmission delay change is calculated based on the link mutation risk value, resource status interval identifier, and link delay change sequence in the vehicle communication status matrix, and the predicted transmission duration is generated based on the link transmission delay change. The predicted transmission completion time is calculated based on the data generation timestamp and the predicted transmission duration. Perform value range segmentation processing on the vehicle data value tags to generate data value range identifiers for the corresponding vehicle business data; Perform time offset correlation processing based on the time offset between the predicted transmission completion time and the cloud receiving window to generate a window time offset sequence; Window state encoding is performed based on the window time series offset sequence, data value interval identifier, and risk interval identifier to generate a data window associated state vector. Perform window matching interval partitioning on the associated state vector of the data window to generate the data window matching result and the corresponding matching priority identifier.

[0029] In this implementation, the cloud receiving window is divided into multiple window periods according to a continuous time interval, and each window period corresponds to a different communication scheduling interval identifier; the link transmission delay change is generated based on the change amplitude of the link delay change sequence between continuous acquisition times, and the predicted transmission duration is jointly generated based on the link transmission delay change and the resource occupancy status corresponding to the resource status interval identifier; the window time sequence offset sequence is generated according to the time difference between the predicted transmission completion time and the cloud receiving window start time; the data window associated state vector is generated according to the correspondence between the window time sequence offset sequence, the data value interval identifier, and the risk interval identifier by performing dimensional mapping; and the matching priority identifier is generated according to the offset amplitude interval in the window time sequence offset sequence.

[0030] In this embodiment, step six includes: Extract vehicle business data from vehicle communication associated data, and perform value range classification processing on vehicle business data based on vehicle data value tags to generate high-value data range, medium-value data range and low-value data range; Perform field association splitting processing on vehicle business data, and divide key status fields and extended status fields based on the temporal association and transmission dependency between data fields; Core slices are generated based on key status fields, supplementary slices are generated based on extended status fields, and slice association indexes are established between core slices and supplementary slices according to the data generation timestamp. Extract the matching priority identifier, scheduling interval identifier, and risk interval identifier from the data window matching results, and perform status association sorting processing according to the unified time index; Based on the link mutation risk value, matching priority identifier, scheduling interval identifier, and risk interval identifier, action status encoding processing is performed to generate a sending action status vector. Perform action interval partitioning processing on the sending action state vector to generate the sending action type corresponding to the vehicle service data; Based on the sending action type, the core slice and supplementary slice are processed to perform action association mapping, and the corresponding vehicle service data sending action is generated.

[0031] In this implementation, key status fields include vehicle control status fields, emergency event status fields, and vehicle cooperative interaction status fields; extended status fields include environmental perception fields, historical status fields, and redundant description fields; the slice association index is generated according to the data generation timestamps and field association order corresponding to the core slice and supplementary slice; the sending action status vector is generated by performing dimensional mapping and arrangement according to the correspondence between link mutation risk value, matching priority identifier, scheduling interval identifier, and risk interval identifier; the sending action types include core slice priority sending type, core slice and supplementary slice joint sending type, supplementary slice delayed sending type, and supplementary slice restricted sending type; the action association mapping processing includes performing corresponding transmission order arrangement processing on the core slice and supplementary slice based on the sending action type.

[0032] In this embodiment, step seven includes: Extract the sending actions corresponding to vehicle service data, and perform transmission order sorting processing on the core slice and supplementary slice based on the sending action type; Establish a corresponding data transmission queue based on the sending action type, and perform queue writing processing on the core slice and supplementary slice according to the data generation timestamp; Perform transmission status recording processing on the core slices and supplementary slices in the data transmission queue, recording the upload completion time, retransmission count, queue waiting time, and cloud reception status of the corresponding slices; A transmission delay sequence is generated based on the time difference between the upload completion time and the data generation timestamp, and a transmission stability sequence is generated based on the number of retransmissions and the cloud reception status. Based on the transmission delay sequence, transmission stability sequence, and cloud reception status, perform reception validity encoding processing to generate historical reception validity data; Perform time-interval statistical processing on historical reception validity data to generate a historical validity change sequence; Based on the historical validity change sequence, the resource status interval identifier and risk interval identifier in the regional communication intent code, the interval status update process is performed to generate the updated regional communication intent code and send it to the corresponding roadside edge node.

[0033] In this implementation, data transmission queues are divided into core slice priority transmission queues, joint transmission queues, and restricted transmission queues according to the type of transmission action. Different data transmission queues correspond to different queue scheduling orders. The transmission stability sequence is generated according to the correspondence between the change in the number of retransmissions in a continuous time interval and the cloud reception status. Historical reception validity data is generated according to the time correlation between the transmission delay sequence, the transmission stability sequence, and the cloud reception status. The historical validity change sequence is generated according to the change trend of historical reception validity data in a continuous time interval. During the update of the regional communication intent code, the resource status interval identifier and the risk interval identifier are synchronously updated according to the change interval corresponding to the historical validity change sequence.

[0034] refer to Figure 2 A cloud-based collaborative vehicle data communication system includes: The data acquisition and value labeling module is used to collect vehicle communication-related data and data generation timestamps, and generate vehicle data value tags. The state matrix construction module is used to construct the vehicle communication state matrix based on the communication link data in the vehicle communication associated data. The link risk identification module is used to construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate link mutation risk values ​​based on the sparse coefficient matrix corresponding to the current link stage. The regional intent code generation module is used to generate regional communication intent codes based on link mutation risk values ​​and vehicle communication association data, and then distribute them to roadside edge nodes. The window matching analysis module is used to generate data window matching results based on the link mutation risk value, vehicle data value label, predicted transmission completion time, and cloud receiving window in the regional communication intent code. The slice scheduling decision module is used to perform hierarchical slicing processing on vehicle business data in vehicle communication associated data based on vehicle data value tags, generate core slices and supplementary slices, and determine the sending action corresponding to vehicle business data by combining link mutation risk value, data window matching result and regional communication intent code. The transmission feedback update module is used to perform vehicle business data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

[0035] Example 1: To verify the feasibility of this invention in practice, it was applied to a vehicle-to-everything (V2X) cooperative communication scenario on an urban expressway. The test area was a section of road from the city center to the airport expressway, approximately 32 kilometers long, including elevated bridges, ramps, tunnels, and high-density traffic light intersections. Twelve roadside edge nodes and one regional cloud dispatch server were deployed within the section, connecting a total of 286 intelligent vehicles with V2X communication capabilities. During vehicle operation, autonomous driving control data, vehicle cooperative interaction data, environmental perception data, and high-frequency status synchronization data were continuously generated. The vehicle terminals interacted with the roadside edge nodes via cellular V2X links.

[0036] In actual operation, due to high-speed vehicle movement, fluctuating edge node load, and frequent road emergencies, traditional vehicle communication scheduling methods are prone to problems such as drastic fluctuations in link status, failures in critical business data transmission, and uneven edge resource utilization. For example, during the evening rush hour, a large number of vehicles gather near airport expressway toll stations in a short period of time, causing the edge node buffer queue length to continuously increase, resulting in a significant increase in the upload delay of vehicle control status data in traditional fixed-priority communication methods. At the same time, in tunnel sections and elevated switching areas, due to rapid changes in communication signals, vehicle communication links are prone to instantaneous changes, causing some vehicle collaborative interaction data to be uploaded only after the cloud receiving window closes, resulting in data failure. To address the above problems, this invention constructs a vehicle communication state matrix to jointly model changes in signal-to-noise ratio, link latency, packet loss, and edge buffer queue in the vehicle link state, and uses the stage inheritance residual triggered switching SINDy model to identify the dynamic evolution patterns in the vehicle communication link.

[0037] During vehicle operation, the system continuously collects vehicle business data, vehicle status data, communication link data, and road event data, and performs unified time-series alignment processing based on the data generation timestamps. Subsequently, the system performs value interval segmentation processing on the vehicle business data, classifying autonomous driving control commands, emergency braking states, and cooperative obstacle avoidance information into high-value data intervals, and environmental perception images and historical trajectory data into medium- and low-value data intervals. The system further constructs a candidate function library based on the vehicle communication state matrix, and dynamically adjusts the sparse competition relationship between candidate function terms through an action-sensitive sparse competition structure to achieve sparse identification of link state change characteristics under different communication behavior conditions.

[0038] When a vehicle enters a high-density congestion area, the regional communication intent code generation module generates a corresponding communication scheduling interval identifier based on the regional vehicle density change sequence, resource pressure change sequence, and link mutation risk value, and sends the regional communication intent code to the roadside edge nodes. The vehicle terminal dynamically adjusts its data transmission actions according to the cloud receiving window in the regional communication intent code. For vehicle control status data in high-value data intervals, core slices are generated first and uploaded preferentially; for environmental perception extended data, supplementary slices are generated and sent with a delay. When a continuously increasing link mutation risk value is detected, the system automatically limits the proportion of supplementary slices sent, prioritizing the uploading of critical control data within the cloud receiving window.

[0039] This invention models the dynamic evolution of vehicle communication links using a stage-inherited residual-triggered switching SINDy model. This model can identify sudden changes in link states in advance and, combined with regional communication intent codes, data window matching mechanisms, and vehicle service data hierarchical slicing mechanisms, enables dynamic collaborative communication control between vehicle terminals, roadside edge nodes, and cloud servers. In high-density vehicle communication environments, this invention significantly reduces the latency of critical service data transmission, improves the success rate of cloud receiving window matching, and reduces edge node resource occupancy. Furthermore, this invention utilizes historical reception validity data to dynamically update regional communication intent codes, continuously adapting to changes in vehicle communication link states and improving communication stability and the reliability of critical service data transmission in complex road scenarios.

[0040] In actual testing, this invention was compared with traditional fixed-priority communication methods and ordinary dynamic queue scheduling methods. The test period was 7 consecutive days, with approximately 3.6TB of vehicle communication data collected each day. Indicators such as vehicle business data upload success rate, average transmission latency of key business data, link mutation identification accuracy, and edge node resource occupancy rate were statistically analyzed. The experimental results are shown in Table 1.

[0041] Table 1 Comparison of Test Results for Vehicle-to-Everything (V2X) Collaborative Communication

[0042] As shown in Table 1, this invention demonstrates high communication stability and link scheduling capabilities in complex vehicle-to-everything (V2X) communication scenarios. Compared to fixed-priority communication methods and dynamic queue scheduling methods, this invention significantly improves indicators such as the success rate of critical business data upload, the accuracy of link mutation identification, and the success rate of cloud receiving window matching. The success rate of critical business data upload reaches 98.1%, indicating that this invention can effectively guarantee the transmission of critical vehicle business data in environments with high congestion and link fluctuations. Simultaneously, the average transmission latency is reduced to 81ms, and the average number of retransmissions is reduced to 1.4, demonstrating that this invention effectively reduces the problem of repeated transmissions caused by link fluctuations through the stage inheritance residual-triggered switching SINDy model and data window matching mechanism. Furthermore, the average resource utilization rate of edge nodes decreases to 64.9%, indicating that this invention can achieve dynamic collaborative scheduling of edge resources and improve the overall efficiency of communication resource utilization.

[0043] The above description is only 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 vehicle data communication method based on cloud collaboration, characterized in that, Includes the following steps: Step 1: Collect vehicle communication-related data and data generation timestamps to generate vehicle data value tags; Step 2: Construct a vehicle communication state matrix based on the communication link data in the vehicle communication association data; Step 3: Construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate link mutation risk values ​​based on the sparse coefficient matrices corresponding to adjacent link stages. Step 4: Generate a regional communication intent code based on the link mutation risk value and vehicle communication association data, and send it to the roadside edge node; Step 5: Generate a data window matching result based on the link mutation risk value, vehicle data value tag, predicted transmission completion time, and cloud receiving window in the regional communication intent code; Step 6: Based on the vehicle data value tag, perform hierarchical slicing processing on the vehicle business data in the vehicle communication association data to generate core slices and supplementary slices, and determine the sending action corresponding to the vehicle business data by combining the link mutation risk value, data window matching result and regional communication intent code. Step 7: Perform vehicle service data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

2. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step one includes: It collects vehicle business data, vehicle status data, communication link data, regional vehicle density data, edge node load data, cloud queue data, and road event data during vehicle operation, and generates timestamps based on the collection time. The vehicle business data is divided into three categories: safety event data, status synchronization data, and background awareness data. Corresponding data type identifiers are then established based on these categories. Perform state change detection processing on vehicle state data, calculate the changes in speed, acceleration, and steering angle between adjacent acquisition times, and generate a vehicle dynamic state identifier based on the state change. Perform link timing processing on the communication link data, and arrange the signal-to-noise ratio, round-trip delay, number of retransmissions and packet loss rate in time series according to the data generation timestamp to form a communication link state sequence; Spatial clustering is performed on regional vehicle density data to generate regional vehicle density distribution results based on vehicle location coordinates and distances between vehicles. Interval statistical processing is also performed on edge node load data and cloud queue data according to time series to form an edge resource status sequence. The road event data is processed for event association. Based on the location of the road event, the duration of the event and the road segment associated with the event, a road event impact identifier is generated, and time sequence alignment is performed according to the data generation timestamp and the communication link status sequence. Based on data type identifiers, vehicle dynamic status identifiers, communication link status sequences, regional vehicle density distribution results, edge resource status sequences, and road event impact identifiers, state association coding processing is performed to generate vehicle data value tags.

3. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step two includes: Extract communication link data from vehicle communication association data, and perform time-series sorting processing on the communication link data according to the data generation timestamp; Based on the time-series sorted communication link data, the signal-to-noise ratio, round-trip delay, number of retransmissions, packet loss rate, and edge node buffer queue length are extracted, and data alignment processing is performed according to the collection time. The signal-to-noise ratio (SNR) is divided into fluctuation intervals, the SNR change between adjacent acquisition times is calculated, and an SNR change sequence is generated. Perform time series correlation permutation processing on round-trip delay, retransmission count, and packet loss rate to generate link delay change sequence, retransmission change sequence, and packet loss change sequence; The relative displacement between vehicles is calculated based on the vehicle's position coordinates and the position coordinates of adjacent vehicles, and a relative speed sequence of adjacent vehicles is generated based on the relative displacement between vehicles. Perform time-series permutation processing on the length of the edge node cache queue to generate a change sequence of the edge cache queue; Based on the signal-to-noise ratio change sequence, link delay change sequence, retransmission change sequence, packet loss change sequence, relative speed sequence of neighboring vehicles, and edge buffer queue change sequence, a vehicle communication state matrix is ​​generated by performing vector combination processing according to a unified time index.

4. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step three includes: A candidate function library is constructed based on the vehicle communication state matrix. The candidate function library includes signal-to-noise ratio variation items, link delay variation items, retransmission variation items, packet loss variation items, relative speed variation items of neighboring vehicles, edge buffer queue variation items, as well as product coupling items and timing differential coupling items between various state variables. The candidate function library is input into the improved SINDy model of stage inheritance residual triggering switching. The improved SINDy model of stage inheritance residual triggering switching includes a stage sparse identification structure, a residual triggering switching structure, a stage inheritance sparse structure, and an action-sensitive sparse competition structure. The action-sensitive sparse competition structure is used to adjust the sparse competition relationship between candidate function items based on the historical communication action type. The candidate function library is partitioned into candidate groups using the aforementioned action-sensitive sparse competition structure, including: Based on the historical communication action types in the vehicle communication association data, the candidate function items are divided into upload association candidate group, relay association candidate group, cache association candidate group and slice association candidate group; Based on the candidate group corresponding to the current historical communication action type, the sparse competition order between candidate function items is adjusted, and sparse coefficient compression processing is performed on the adjusted candidate function items to generate a set of action-related candidate functions. The sparse dynamics identification process is performed on the action association candidate function set using the stage sparse identification structure. Based on the candidate function terms, the sparse coefficients are solved for the dynamic evolution relationship of the state variables in the vehicle communication state matrix to generate the sparse coefficient matrix corresponding to the current link stage. The residual-triggered switching structure is used to perform dynamic residual detection processing on the sparse coefficient matrix corresponding to the current link stage, including: The state prediction result is calculated based on the sparse coefficient matrix corresponding to the current link stage. Calculate the state residual sequence based on the state prediction results and the actual state values ​​in the vehicle communication state matrix; Perform time window accumulation processing on the state residual sequence to generate a residual change sequence; Based on the growth segment of the residual change sequence in the continuous time interval, perform link stage switching detection and generate a new link stage identifier; The stage inheritance sparse structure is used to perform stage inheritance constraint processing on the sparse coefficient matrix corresponding to the new link stage, including: Extract the non-zero sparse coefficients corresponding to the current link stage; Based on the non-zero sparse coefficient terms, perform sparse constraint update processing on the candidate function terms in the new link stage; Based on the updated candidate function terms, the sparse coefficient solution process is re-executed to generate the sparse coefficient matrix corresponding to the new link stage. Sparse difference calculation is performed on the sparse coefficient matrix corresponding to adjacent link stages to generate a sparse change matrix. Based on the non-zero change terms in the sparse change matrix, the state offset between link stages is calculated to generate a link mutation risk value.

5. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step four includes: Extract regional vehicle density data, edge node load data, cloud queue data, and road event data from vehicle communication association data, and perform time-series association and sorting processing according to the data generation timestamp; Spatial aggregation and correlation processing is performed on regional vehicle density data to generate a vehicle density change sequence based on the distance change relationship between vehicle location coordinates. Perform resource time-series correlation processing on edge node load data and cloud queue data, and generate a resource pressure change sequence based on the changes in edge node load and cloud queue length within a continuous time interval; Perform event impact correlation processing on road event data to generate road event impact sequences based on the location, duration, and associated road segments of the road events. Based on the link mutation risk value, vehicle density change sequence, resource pressure change sequence, and road event impact sequence, multi-dimensional state coding processing is performed to generate a regional communication state vector. Perform state interval partitioning on the regional communication state vector to generate communication scheduling interval identifiers for the corresponding state intervals; Based on the communication scheduling interval identifier, link mutation risk value, and resource pressure change sequence, interval coding processing is performed to generate a regional communication intent code including the scheduling interval identifier, risk interval identifier, and resource status interval identifier, and the regional communication intent code is sent to the corresponding roadside edge node.

6. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step five includes: Extract the scheduling interval identifier, risk interval identifier, and resource status interval identifier from the regional communication intent code, and extract the start and end times of the corresponding cloud receiving window; The link transmission delay change is calculated based on the link mutation risk value, resource status interval identifier, and link delay change sequence in the vehicle communication status matrix, and a predicted transmission duration is generated based on the link transmission delay change. The predicted transmission completion time is calculated based on the data generation timestamp and the predicted transmission duration. Perform value range segmentation processing on the vehicle data value tags to generate data value range identifiers for the corresponding vehicle business data; Perform time offset correlation processing based on the time offset between the predicted transmission completion time and the cloud receiving window to generate a window time offset sequence; Window state encoding is performed based on the window time series offset sequence, data value interval identifier, and risk interval identifier to generate a data window associated state vector. Perform window matching interval partitioning processing on the data window associated state vector to generate data window matching results and corresponding matching priority identifiers.

7. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step six includes: Extract vehicle business data from vehicle communication associated data, and perform value range classification processing on vehicle business data based on vehicle data value tags to generate high-value data range, medium-value data range and low-value data range; Perform field association splitting processing on vehicle business data, and divide key status fields and extended status fields based on the temporal association and transmission dependency between data fields; Core slices are generated based on key status fields, supplementary slices are generated based on extended status fields, and slice association indexes are established between core slices and supplementary slices according to the data generation timestamp. Extract the matching priority identifier, scheduling interval identifier, and risk interval identifier from the data window matching results, and perform status association sorting processing according to the unified time index; Based on the link mutation risk value, matching priority identifier, scheduling interval identifier, and risk interval identifier, action status encoding processing is performed to generate a sending action status vector. Perform action interval partitioning processing on the sending action state vector to generate the sending action type corresponding to the vehicle service data; Based on the sending action type, the core slice and supplementary slice are processed to perform action association mapping, and the corresponding vehicle service data sending action is generated.

8. The vehicle data communication method based on cloud collaboration according to claim 1, characterized in that, Step seven includes: Extract the sending actions corresponding to vehicle service data, and perform transmission order sorting processing on the core slice and supplementary slice based on the sending action type; Establish a corresponding data transmission queue based on the sending action type, and perform queue writing processing on the core slice and supplementary slice according to the data generation timestamp; Perform transmission status recording processing on the core slices and supplementary slices in the data transmission queue, recording the upload completion time, retransmission count, queue waiting time, and cloud reception status of the corresponding slices; A transmission delay sequence is generated based on the time difference between the upload completion time and the data generation timestamp, and a transmission stability sequence is generated based on the number of retransmissions and the cloud reception status. Based on the transmission delay sequence, transmission stability sequence, and cloud reception status, perform reception validity encoding processing to generate historical reception validity data; Perform time-interval statistical processing on historical reception validity data to generate a historical validity change sequence; Based on the historical validity change sequence, the resource status interval identifier and risk interval identifier in the regional communication intent code, the interval status update process is performed to generate the updated regional communication intent code and send it to the corresponding roadside edge node.

9. A cloud-based collaborative vehicle data communication system, comprising executing the cloud-based collaborative vehicle data communication method according to any one of claims 1 to 8, characterized in that, include: The data acquisition and value labeling module is used to collect vehicle communication-related data and data generation timestamps, and generate vehicle data value tags. The state matrix construction module is used to construct the vehicle communication state matrix based on the communication link data in the vehicle communication associated data. The link risk identification module is used to construct a candidate function library based on the vehicle communication state matrix, perform segmented sparse identification processing on the candidate function library using the improved SINDy model, generate sparse coefficient matrices corresponding to multiple link stages, and generate link mutation risk values ​​based on the sparse coefficient matrices corresponding to adjacent link stages. The regional intent code generation module is used to generate regional communication intent codes based on link mutation risk values ​​and vehicle communication association data, and then distribute them to roadside edge nodes. The window matching analysis module is used to generate data window matching results based on the link mutation risk value, vehicle data value label, predicted transmission completion time, and cloud receiving window in the regional communication intent code. The slice scheduling decision module is used to perform hierarchical slicing processing on vehicle business data in vehicle communication associated data based on vehicle data value tags, generate core slices and supplementary slices, and determine the sending action corresponding to vehicle business data by combining link mutation risk value, data window matching result and regional communication intent code. The transmission feedback update module is used to perform vehicle business data transmission processing based on the sending action, generate historical reception validity data, and update the regional communication intent code based on the historical reception validity data.

Images