A driving safety warning system and method based on Internet of Vehicles

By building a quantum-enhanced high-precision digital twin model and deep learning-driven data fusion technology in the Internet of Vehicles system, combined with edge computing, the problem of insufficient real-time and accuracy of the Internet of Vehicles system in driving safety warning is solved, and the improvement is achieved Fast and precise risk assessment and response strategy selection significantly improves the safety and synergistic efficiency of the transportation network.

CN118470974BActive Publication Date: 2024-11-05GUANGXI BOLIAN NEW ENERGY TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410890666.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-04
Publication Date
2024-11-05
Estimated Expiration
2044-07-04

AI Technical Summary

Technical Problem

The existing Internet of Vehicles systems have problems such as insufficient real-time and accuracy in driving safety warning, low efficiency in sensor data fusion processing, challenges in the application of multimodal data fusion in dynamic traffic environments, and underexploited quantum computing potential.

Method used

By collecting vehicle surrounding environment data in real time, using the on-board vehicle network communication architecture to communicate between vehicles and vehicles and vehicles with infrastructure, building a quantum-enhanced high-precision digital twin model, and using deep learning-driven data fusion algorithm to interpret the vehicle's own perceived data and Multiple external information is fused in space-time multimodal data, and adaptive dynamic risk assessment results are generated, and risk situation simulation is carried out through quantum computing to form optimal risk response measures, and early warning signals are generated in real time with edge computing technology.

Benefits of technology

It significantly improves the speed and accuracy of data processing, realizes more personalized and accurate early warning services, shortens emergency response time, and improves the coordination efficiency and security of the transportation network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118470974B_ABST
    Figure CN118470974B_ABST
Patent Text Reader

Abstract

The present invention discloses a driving safety warning system and method based on the Internet of Vehicles, which relates to the technical field of intelligent transportation systems, including real-time collection of vehicle surrounding environment data, communication between vehicles and between vehicles and infrastructure through the vehicle-mounted Internet of Vehicles communication architecture, and generation of vehicle-sensing data; using edge computing technology, executing warning decision algorithms on the edge data nodes of the Internet of Vehicles, and generating warning signals in real time based on the optimal risk response measures, and accurately positioning and immediately transmitting them to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow through the Internet of Vehicles platform. The present invention optimizes timestamp matching and deepens multimodal data fusion, combined with innovative quantum computing applications, to build a more real-time and accurate Internet of Vehicles driving safety warning system, significantly improves the adaptive assessment and response capabilities for risks in complex traffic environments, and promotes the advancement of intelligent management of traffic safety.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent transportation systems, and in particular to a driving safety warning system and method based on Internet of Vehicles. Background Art

[0002] In today's society, with the rapid development of information technologies such as the Internet of Things, big data, cloud computing, and artificial intelligence, the Internet of Vehicles has become a core component of the intelligent transportation system. It has greatly improved the safety and efficiency of road transportation through all-round communication between vehicles, vehicles and infrastructure, and even vehicles and pedestrians. The application of Internet of Vehicles technology, especially in driving safety warning systems, has gradually demonstrated its great potential in preventing traffic accidents and optimizing traffic flow management. Early warning systems mainly rely on sensors of individual vehicles, such as radars and cameras, to monitor the surrounding environment, but are limited by sensor performance and information processing capabilities, and their warning accuracy and timeliness are limited.

[0003] Although the environmental perception capability of existing Internet of Vehicles systems has been enhanced, its real-time and accuracy still need to be improved; the fusion processing of sensor data is usually simple, and lacks an efficient timestamp matching optimization mechanism, resulting in an untimely response to emergencies and reducing the reliability of the early warning system; secondly, although deep learning has achieved remarkable results in multimodal data fusion, how to accurately integrate the vehicle's own perception data with external multi-dimensional information (such as traffic signals, weather conditions, etc.) in real time in a dynamically changing traffic environment, and generate adaptive dynamic risk assessments based on this, is still a major technical problem; in addition, the theoretical advantages of quantum computing have not yet been fully demonstrated in practical applications, especially in risk assessment and response strategy selection in complex traffic scenarios, there is still a lack of mature methodology and implementation cases. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a driving safety warning system and method based on the Internet of Vehicles to solve the problems in the prior art of insufficient real-time and accuracy, low efficiency of sensor data fusion processing, application challenges of multimodal data fusion in dynamic traffic environments, and insufficient exploration of the potential of quantum computing.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In the first aspect, an embodiment of the present invention provides a driving safety warning method based on the Internet of Vehicles, which includes: real-time collection of vehicle surrounding environment data, communication between vehicles and between vehicles and infrastructure through the on-board Internet of Vehicles communication architecture, and generation of vehicle-sensing data; timestamp matching optimization of the vehicle surrounding environment data collected in real time, and construction of a quantum-enhanced high-precision digital twin model; using a deep learning-driven data fusion algorithm to perform spatiotemporal multimodal data fusion of the vehicle's own perception data with multivariate external information received through the on-board Internet of Vehicles, and generating an adaptive dynamic risk assessment result; through the quantum-enhanced high-precision digital twin model, risk scenario simulation of the adaptive dynamic risk assessment result is performed to form an optimal risk response measure; using edge computing technology, an early warning decision algorithm is executed on the edge data node of the Internet of Vehicles, and an early warning signal is generated in real time according to the optimal risk response measure, and the early warning signal is accurately located through the Internet of Vehicles platform and immediately transmitted to the vehicle closest to the risk area and the infrastructure node responsible for regulating traffic flow.

[0008] As a preferred solution of the driving safety warning method based on the Internet of Vehicles of the present invention, the real-time collection of vehicle surrounding environment data, the communication between vehicles and vehicles and vehicles and infrastructure through the vehicle-mounted Internet of Vehicles communication architecture, and the generation of vehicle-sensing data, the specific steps are:

[0009] Various sensors such as high dynamic range night vision cameras, millimeter wave radars, lidars, and ultrasonic sensors are installed at key locations on the front and rear bumpers, under the side mirrors, and on the roof of the vehicle;

[0010] The various sensors installed collect data about the vehicle's surrounding environment, such as images, distance, and speed, and use a timestamp synchronization mechanism to output initial sensor data;

[0011] The Kalman filter algorithm is used to remove noise from the initial sensor data, and the background subtraction and edge detection image processing algorithms are used to optimize the quality of the camera image data. Cluster analysis is performed through the radar point cloud to distinguish between static and dynamic objects, and then integrated to form standard driving sensor data;

[0012] Aggregate standard driving sensor data from the vehicle bus to a central processor for image frames, radar point clouds, ultrasonic echoes and vehicle posture information;

[0013] The camera images and radar point cloud data from standard driving sensor data are clustered and analyzed, combined with machine vision and YOLOv5 deep learning algorithm to build a 3D environment model.

[0014] The integrated communication protocol supporting V2V and V2I will be connected to the Internet of Vehicles, integrating the three-dimensional environmental model and standard driving sensor data, and communicating between vehicles and between vehicles and infrastructure through the on-board communication architecture to generate the vehicle's own perception data.

[0015] As a preferred solution of the driving safety warning method based on the Internet of Vehicles of the present invention, the specific steps of performing timestamp matching optimization on the real-time collected vehicle surrounding environment data and constructing a quantum enhanced high-precision digital twin model are as follows:

[0016] Based on the driving sensor data, quantum entanglement is used to perform preprocessing and timestamp matching optimization. The expression is: ; Among them, |Ψ + > is the quantum entangled state, |00> and |11> represent the states of the first and second quantum bits, respectively;

[0017] Based on the exchange mechanism in the quantum entangled state, the timestamps are matched and the synchronized data is represented as the quantum state |Ψ sc >

[0018] Using quantum feature mapping, the synchronized quantum state |Ψ sc >Mapped to the feature space, the quantum state expression after feature extraction is: ; Among them, |Φ ft > is the quantum state after feature extraction, U ft Quantum operations for feature mapping;

[0019] The expression for training a set of quantum states of the support vector machine based on the quantum state after feature extraction is: ; Among them, |Φ tn,i > is the quantum state of the i-th training sample, and M is the number of samples in the training set;

[0020] Based on the trained quantum state, the similarity expression is calculated by quantum inner product: ;in, For the test state |Φ tt >with the quantum state of the i-th training sample |Φ tn,i > the similarity between is the feature mapping quantum operation U ft The transposed conjugate of

[0021] Substitute the similarity value calculated by quantum inner product into the decision rule of quantum support vector machine, and use quantum version of Lagrange multiplier method and kernel technique to make prediction expression of quantum support vector machine: ; Among them, class (|Φ tt>) represents the classification decision result of the test state based on the quantum support vector machine, sign is the sign function that determines the positive or negative classification result, α i is the Lagrange multiplier of the i-th training sample, yi is the label of the i-th training sample, and b is the bias term for adjusting the position of the classification hyperplane;

[0022] The classification decision result of the test state based on the quantum support vector machine is used as a classification indication of the current vehicle environment state and integrated into the dynamic update mechanism of the quantum digital twin model to construct a quantum enhanced high-precision digital twin model. The expression is: ; where t represents the time step, |Ψ QDT (t)> is the quantum state of the vehicle environment at time t, U ee It is the evolution operator that directly reflects the quantum support vector classification result, and U is the operator.

[0023] As a preferred solution of the driving safety warning method based on the Internet of Vehicles of the present invention, the deep learning-driven data fusion algorithm is used to perform spatiotemporal multimodal data fusion on the vehicle's own perception data and the multivariate external information received through the vehicle-mounted Internet of Vehicles to generate an adaptive dynamic risk assessment result. The specific steps are:

[0024] Receive multiple external information such as traffic signal status, weather forecast, road construction information, vehicle location and speed through the vehicle-mounted Internet of Vehicles;

[0025] Based on the vehicle's own perception data and the multivariate external information received by the vehicle-mounted Internet of Vehicles, the GPS positioning data and timestamp matching method are combined for time synchronization and spatial registration to form fused data;

[0026] For the image data and radar data in the vehicle's own perception data, deep convolutional neural networks and point cloud processing algorithms are used to extract high-level semantics and multi-modal features of the object's motion trajectory;

[0027] The spatiotemporal attention mechanism STAM is used to fuse the extracted multimodal features: Among them, F fn is the fused feature vector, F ie , F rr , ..., F el Represent the feature vectors extracted from images, radars, external information, etc. respectively; based on the fused feature vectors, a deep neural network model is constructed; the input layer of the deep neural network model accepts the fused feature vector expression as follows: ; ; Among them, a (l) For the The activation output of the layer, f (l) For the The ReLU activation function used by the layer, z (l) is the weighted input of the lth layer, W (l) For the -1st floor to The weight matrix of the layer, b (l) For the The bias vector of the layer;

[0028] For the output layer of the deep neural network model, the risk level expression is obtained through the softmax function: ; Among them, P(y=k|F fn ) is the feature vector F after fusion fn Under the condition of , the probability that the true risk level label y belongs to the category label k, e zk is the linear weighted input value z of the output node for category k, C is the total number of categories, and j is the category index;

[0029] The cross entropy loss function is selected as the target of supervised learning of the deep neural network model, and the Adam optimizer is used to update the model parameters. The loss function model L(y, ) is expressed as: ; Among them, y o,c is the value of the true risk level label y under sample o and category c, The probability distribution predicted by the loss function model;

[0030] After the cross entropy loss function is performed, the optimization fitting of the deep neural network model is completed, and the adaptive dynamic risk assessment results are generated by receiving the real-time input of the vehicle's own perception data and the multivariate external information received by the on-board Internet of Vehicles.

[0031] As a preferred solution of the driving safety warning method based on the Internet of Vehicles described in the present invention, wherein: the risk scenario simulation is performed on the adaptive dynamic risk assessment results through the quantum enhanced high-precision digital twin model to form the optimal risk response measures, and the specific steps are:

[0032] Each risk level in the adaptive dynamic risk assessment result is mapped to a quantum state, and the quantum risk scenario space expression is constructed by converting binary into quantum bit sequence: ; Among them, | > represents the state of the i-th risk level, |j> is the quantum ground state, n represents the quantum resource amount encoding the risk level information, a is the amplitude coefficient, i and j are the subscripts of the risk level and quantum ground state respectively;

[0033] Design of quantum risk evolution operator U el , the expression of the simulated risk situation space changing over time is: Among them, U el is the quantum operator that controls the evolution of risk situation, △t is the time step, and θ is the environmental variable control parameter;

[0034] By utilizing the parallel processing capability of quantum computers, we simulate the vehicle behavior response under various risk scenarios through quantum evolution to form a super position state |Ψ sim The expression is: ; Where R is the total number of risk levels;

[0035] Based on the formation of a series of quantum states of super-position states, information is extracted from the super-position states through quantum measurement, and the effect expression of quantifying different risk response strategies is: ; Where E(O) is the expected value function, and O is the observation operator of the physical quantity for evaluating the effectiveness of risk response strategies;

[0036] Combining the quantum evolution simulation and quantum measurement results, the risk response measures with the highest expected benefits are selected, expressed as: ; Among them, sy ol is the optimal risk response strategy selected through the optimization process. It refers to finding the value s at which the expected value function reaches the maximum value among all risk response strategies.

[0037] As a preferred solution of the driving safety warning method based on the Internet of Vehicles of the present invention, the edge computing technology is used to execute the warning decision on the edge data node of the Internet of Vehicles, and the warning signal is generated immediately according to the optimal risk response measures. The specific steps are:

[0038] Deploy edge computing devices and efficient data processing and transmission middleware at data nodes at edge traffic fortresses, city entrances, and accident-prone areas of the Internet of Vehicles;

[0039] The collected data resources of edge computing nodes are optimized using an adaptive dynamic risk assessment model combined with a quantum enhancement algorithm;

[0040] When the quantum enhanced algorithm optimization analysis shows that the probability of the risk state exceeds the preset threshold, an early warning is immediately triggered and an early warning decision is deployed.

[0041] As a preferred solution of the driving safety warning method based on the Internet of Vehicles of the present invention, the precise positioning through the Internet of Vehicles platform is immediately transmitted to the vehicle closest to the risk area and the infrastructure node responsible for regulating the traffic flow, and the specific steps are:

[0042] With the help of the vehicle's own perception data and multiple external information, it is transmitted to the nearest edge computing node in real time through the V2X communication network, and the received data is initially cleaned, deduplicated and formatted;

[0043] Input the pre-processed data into the deployed early warning decision-making, and use the quantum-enhanced high-precision digital twin model to quickly determine the risk level of the vehicle's environment;

[0044] According to the risk level of the environment in which the vehicle is located, the edge node instantly extracts information from the super position state, calculates the expected benefits of different response strategies through quantum measurement, and selects the optimal risk response strategy with the highest expected benefit as a guide for action;

[0045] Based on the selected action guidelines, the edge node instantly generates visual warnings, sound alarms, and transmits the warning information to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow through the V2V communication network.

[0046] In the second aspect, the present invention provides a driving safety warning system based on the Internet of Vehicles, including a perception and communication module, a data processing and fusion module, an intelligent analysis and decision-making module, an edge computing and warning module, and an integrated control and interaction module; the perception and communication module is used to collect vehicle surrounding environment data in real time, and to communicate between vehicles and between vehicles and infrastructure through the on-board Internet of Vehicles communication architecture to generate vehicle perception data; the data processing and fusion module is used to optimize the timestamp matching of the vehicle surrounding environment data collected in real time, and to construct a quantum enhanced high-precision digital twin model; the intelligent analysis and decision-making module is used to use deep learning to drive data fusion The edge computing and early warning module is used to simulate the risk scenarios of the adaptive dynamic risk assessment results through the quantum-enhanced high-precision digital twin model, and form the optimal risk response measures; the integrated control and interaction module is used to use the edge computing technology to execute the early warning decision algorithm on the edge data nodes of the Internet of Vehicles, and generate early warning signals in real time according to the optimal risk response measures, which are accurately located through the Internet of Vehicles platform and immediately transmitted to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating the traffic flow.

[0047] In a third aspect, an embodiment of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, any step of the driving safety warning method based on the Internet of Vehicles as described in the first aspect of the present invention is implemented.

[0048] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, any step of the driving safety warning method based on the Internet of Vehicles as described in the first aspect of the present invention is implemented.

[0049] The beneficial effects of the present invention are as follows: This scheme achieves a major breakthrough in the driving safety warning system by integrating the quantum-enhanced digital twin model and the data fusion technology driven by deep learning, targeting the above-mentioned technical bottlenecks; first, the application of quantum technology significantly improves the speed and accuracy of data processing, especially in timestamp matching optimization and risk scenario simulation, using quantum characteristics to achieve ultra-high-speed synchronization and classification, solving the problem of slow response of traditional systems in complex environments; secondly, the integration of deep learning algorithms enables the system to extract key information from multi-dimensional and multi-modal data, adaptively evaluate dynamic risks, and provide more personalized and accurate warning services; in addition, combined with the instant warning decision-making and communication mechanism of edge computing, it ensures the rapid dissemination of risk information, effectively shortens the emergency response time, and improves the collaborative efficiency and safety of the entire transportation network. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0051] Figure 1 This is a flow chart of the driving safety warning method based on the Internet of Vehicles in Example 1.

[0052] Figure 2 This is a flow chart for instantly generating a warning signal based on the optimal risk response measures in Example 1. DETAILED DESCRIPTION

[0053] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0054] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0055] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0056] Example 1, reference Figure 1 and Figure 2, which is the first embodiment of the present invention, and provides a driving safety warning method based on the Internet of Vehicles, comprising the following steps:

[0057] S1. Collect data about the vehicle's surrounding environment in real time, communicate between vehicles and between vehicles and infrastructure through the vehicle-to-vehicle network communication architecture, and generate the vehicle's own perception data.

[0058] Furthermore, various sensors such as high dynamic range night vision cameras, millimeter wave radars, lidars, and ultrasonic sensors are installed at key locations on the front and rear bumpers, under the side mirrors, and on the roof of the vehicle. These sensors cover all-round perception needs from short distance to long distance, from two-dimensional to three-dimensional, and from bright to low-light environments;

[0059] The various sensors installed collect data about the vehicle's surrounding environment, such as images, distance, and speed. The initial sensor data is output using a timestamp synchronization mechanism to ensure that data from different sensors can accurately correspond in time, providing a consistent basis for subsequent processing.

[0060] The Kalman filter algorithm is applied to remove noise from the initial sensor data, effectively eliminate useless signals, improve the purity and reliability of the data, and use background subtraction and edge detection image processing algorithms to optimize the quality of camera image data, enhance the recognizability of target objects, reduce environmental interference, and perform cluster analysis through radar point clouds to distinguish static and dynamic objects. The integration forms standard driving sensor data, which helps to quickly identify moving obstacles and provide a basis for obstacle avoidance strategies;

[0061] It should be noted that "clustering analysis of radar point clouds to distinguish static and dynamic objects" refers to the use of three-dimensional point cloud data of the surrounding environment collected by radar sensors (point cloud is a collection of spatial coordinate points of surrounding objects determined by radar echoes); in point cloud data, each point represents the location information of a reflective surface in the environment, and cluster analysis is a data analysis method used to group similar data points together. In this context, by clustering point cloud data, points belonging to the same entity (such as vehicles, pedestrians, or fixed obstacles) can be clustered together to distinguish different objects;

[0062] Aggregate standard driving sensor data from the vehicle bus to a central processor that integrates image frames, radar point clouds, ultrasonic echoes, and vehicle posture information to form a comprehensive environmental perception data package;

[0063] The camera images and radar point clouds in the standard driving sensor data are clustered and analyzed, combined with machine vision and YOLOv5 deep learning algorithms to build a 3D environment model, accurately locate each perceived object, and predict its dynamic trajectory;

[0064] The integrated V2V and V2I communication protocols are connected to the Internet of Vehicles, enabling vehicles to exchange data efficiently with surrounding vehicles and road infrastructure through the vehicle communication architecture. The three-dimensional environmental model and standard driving sensor data are integrated to communicate between vehicles and between vehicles and infrastructure through the vehicle communication architecture. This not only enables individual vehicles to make decisions based on broader environmental information, but also promotes the collaborative operation of the entire traffic network, improves overall road safety and traffic efficiency, and ultimately integrates and generates vehicle-sensing data.

[0065] It should also be noted that through highly integrated sensing, communication and intelligent analysis technologies, comprehensive, real-time and accurate perception of the vehicle's surrounding environment is achieved, providing strong data support for driving assistance, collision warning, path planning and other functions; it not only improves the safety performance of a single vehicle, but also lays a solid foundation for promoting the construction of smart transportation systems and the development of future autonomous driving technology.

[0066] S2. Perform timestamp matching optimization on the vehicle surrounding environment data collected in real time to build a quantum-enhanced high-precision digital twin model.

[0067] Furthermore, based on the driving sensor data, quantum entanglement is used to perform preprocessing and timestamp matching optimization, and the expression is: ; Among them, |Ψ + > is a quantum entangled state, |00> and |11> represent the states of the first and second quantum bits respectively (i.e., either all are in the 0 state or all are in the 1 state). The non-local characteristics of quantum entanglement provide an extraordinary way for information processing;

[0068] Based on the exchange mechanism in the quantum entangled state, the timestamps are matched and the synchronized data is represented as the quantum state |Ψ sc >, a process that involves converting raw data into quantum states |Ψ sc >, ensuring accurate temporal alignment of data from different sources;

[0069] Using quantum feature mapping, the synchronized quantum state |Ψ sc >Map to feature space to obtain the quantum state after feature extraction. This mapping helps to reveal hidden patterns in the data and improve the performance of classification or prediction. The quantum state after feature extraction |Φ ft The expression is: ; Among them, |Φ ft > is the quantum state after feature extraction, U ft Quantum operations for feature mapping;

[0070] A set of quantum states of the support vector machine is trained according to the quantum state after feature extraction. The basic learning expression of the support vector machine is: ; Among them, |Φ tn,i > is the quantum state of the i-th training sample, and M is the number of samples in the training set;

[0071] Based on the trained quantum state, the similarity is calculated by quantum inner product. This step utilizes the intrinsic mathematical properties of the quantum state and accelerates the computational tasks with higher complexity. The expression is: ;in, For the test state |Φ tt >with the quantum state of the i-th training sample |Φ tn,i > the similarity between is the feature mapping quantum operation U ft The transposed conjugate of

[0072] Furthermore, the value of similarity calculated by quantum inner product is substituted into the decision rule of quantum support vector machine, and the prediction expression of quantum support vector machine is performed by quantum version of Lagrange multiplier method and kernel technique as follows: ; Among them, class (|Φ tt >) represents the classification decision result of the test state based on the quantum support vector machine, sign is the sign function that determines the positive or negative classification result, α i is the Lagrange multiplier of the i-th training sample, yi is the label of the i-th training sample, and b is the bias term for adjusting the position of the classification hyperplane;

[0073] It should be noted that the decision rule of the quantum support vector machine refers to the use of the principle of the quantum support vector machine, through the Lagrange multiplier method and kernel techniques, combined with the training results to calculate the classification label of the test state, which is achieved by solving an optimization problem.

[0074] The classification decision result of the test state based on the quantum support vector machine is used as a classification indication of the current vehicle environment state and integrated into the dynamic update mechanism of the quantum digital twin model to construct a quantum enhanced high-precision digital twin model. The expression is: ; where t represents the time step, |Ψ QDT (t)> is the quantum state of the vehicle environment at time t, U ee It is the evolution operator that directly reflects the quantum support vector classification result, and U is the operator.

[0075] Preferably, the classification results of the quantum support vector machine are integrated into the construction process of the quantum digital twin model. ee The quantum state acting on the current vehicle environment |Ψ QDT(t)>, update the model to the next time step |Ψ based on the classification results QDT (t+1)>, this dynamic update ensures that the model can reflect changes in the vehicle environment in real time and improve the ability to predict and respond to risks; it should also be noted that the above implementation process uses complex quantum computing technology, especially quantum entanglement, feature mapping, and quantum support vector machine methods, to deeply process and optimize the vehicle's surrounding environment data, which not only improves the speed of data processing, but also improves the accuracy and adaptability of the model, providing strong technical support for building an efficient driving safety warning system.

[0076] S3. Use deep learning driven data fusion algorithm to perform spatiotemporal multimodal data fusion of the vehicle’s own perception data and the multivariate external information received through the vehicle-mounted Internet of Vehicles to generate adaptive dynamic risk assessment results.

[0077] Furthermore, multiple external information including but not limited to traffic signal status, weather forecast, road construction information, vehicle location and speed is received through the vehicle-mounted Internet of Vehicles;

[0078] Based on the vehicle's own perception data and the multi-dimensional external information received by the vehicle-mounted Internet of Vehicles (this information supplements the local perception data from a macro level and enhances the understanding of the overall traffic environment), the GPS positioning data and timestamp matching method are combined for time synchronization and spatial registration (ensuring that all data are comparable in the same time and space framework, which is the premise for the effective fusion of multi-source data), and integrated to form fused data;

[0079] It should be noted that the specific implementation steps for time synchronization and spatial registration and integration to form fused data are as follows:

[0080] First, extract timestamps from each data source (including the vehicle's own sensor data, GPS data, and external information received through the Internet of Vehicles). Vehicle sensor data is usually generated by the ECU (electronic control unit) and comes with a timestamp; GPS data naturally comes with a time stamp when it is received; and external information (such as traffic signal status, road condition updates, etc.) should also include the sending time when it is transmitted;

[0081] Choose an accurate and stable clock as the reference clock, usually the GPS clock, because it is highly synchronized with the global standard time and has a small error. The timestamps of all other data sources will be corrected based on this reference clock.

[0082] By comparing the timestamps of each data source with the time of the reference clock, the time deviation between them is calculated. For data sources with obvious deviations, the time offset is applied to compensate and adjust to ensure that the time stamps of all data are as close to the same moment as possible;

[0083] During system operation, time inconsistency may occur again due to various reasons (such as clock drift), so it is necessary to repeat the above steps regularly to implement dynamic time synchronization and maintain the timeliness of data;

[0084] Determine a unified spatial coordinate system, usually using the WGS84 geographic coordinate system or a locally applicable projection coordinate system, to ensure that all geographic location data is expressed in the same coordinate system. The vehicle's GPS data directly provides WGS84 coordinates, but other data sources may use different coordinate systems. These data need to be converted to a unified coordinate system through a coordinate transformation algorithm. For example, the road network data in the external information may be based on local plane coordinates and need to be converted to a coordinate system through known points.

[0085] The vehicle's GPS data directly provides WGS84 coordinates, but other data sources may use different coordinate systems. These data need to be converted to a unified coordinate system through a coordinate transformation algorithm. For example, the road network data in the external information may be based on local plane coordinates and need to be converted to a coordinate system through known points;

[0086] The vehicle's GPS data directly provides WGS84 coordinates, but other data sources may use different coordinate systems. These data need to be converted to a unified coordinate system through a coordinate transformation algorithm. For example, the road network data in the external information may be based on local plane coordinates and need to be converted to a coordinate system through known points.

[0087] Furthermore, for the image data and radar data in the vehicle's own perception data, deep convolutional neural networks and point cloud processing algorithms are used to extract high-level semantics and multimodal features of the object's motion trajectory (this step is to convert the raw data into a form that can be understood by the machine);

[0088] The spatiotemporal attention mechanism STAM is used to fuse the extracted multimodal features: Among them, F fn is the fused feature vector, F ie , F rr , ..., F el They represent the feature vectors extracted from images, radar, external information, etc.

[0089] Preferably, STAM increases the model's attention to key information by assigning different weights in time and space to different features, thereby improving the fusion effect; based on the fused feature vector, a deep neural network model is constructed, whose input layer receives the feature vector F fused by STAM fn, through a series of hidden layers with activation functions (such as ReLU) to perform nonlinear transformations and gradually abstract features; the calculation of each layer follows the standard feedforward neural network formula, and the input layer of the deep neural network model accepts the fused feature vector expression as follows: ; ; Among them, a (l) For the The activation output of the layer, f (l) For the The ReLU activation function used by the layer, z (l) is the weighted input of the lth layer, W (l) For the -1st floor to The weight matrix of the layer, b (l) For the The bias vector of the layer;

[0090] For the output layer of the deep neural network model, the risk level expression is obtained through the softmax function: ; Among them, P(y=k|F fn ) is the feature vector F after fusion fn Under the condition of , the probability that the true risk level label y belongs to the category label k, e zk is the linear weighted input value z of the output node for category k, C is the total number of categories, and j is the category index;

[0091] The cross entropy loss function is selected as the target of supervised learning of the deep neural network model, and the Adam optimizer is used to update the model parameters. The loss function model L(y, ) is expressed as: ; Among them, y o,c is the value of the true risk level label y under sample o and category c, The probability distribution predicted by the loss function model;

[0092] After the cross entropy loss function is completed, the deep neural network model is optimized and fitted. By receiving the real-time input of the vehicle's own perception data and the multivariate external information received by the vehicle-mounted Internet of Vehicles, an adaptive dynamic risk assessment result is generated, which helps the driver to make quick and accurate decision responses, thereby improving driving safety;

[0093] It should be noted that this implementation process achieves a comprehensive and high-precision assessment of the vehicle driving environment through highly integrated deep learning technology and advanced data fusion strategies, providing strong technical support for improving driving safety in the Internet of Vehicles environment.

[0094] S4. Through the quantum-enhanced high-precision digital twin model, the risk scenario simulation of the adaptive dynamic risk assessment results is carried out to form the optimal risk response measures.

[0095] Furthermore, each risk level (e.g., low, medium, and high risk) in the adaptive dynamic risk assessment results is mapped to a quantum state, and the quantum risk scenario space expression is constructed by converting binary into a quantum bit sequence: ; Among them, | > represents the state of the i-th risk level, |j> is the quantum ground state, n represents the quantum resource amount encoding the risk level information, a is the amplitude coefficient, i and j are the subscripts of the risk level and quantum ground state respectively;

[0096] Preferably, through the above mapping, a quantum risk scenario space is constructed, in which each risk level corresponds to a quantum state. This space provides a basis for simulating how different risk levels evolve over time, and provides the possibility for quantum computing to intervene in risk assessment;

[0097] It should be noted that the mapping relationship here is through the amplitude coefficient a ij to determine, each coefficient reflects the probability amplitude of the corresponding quantum ground state |j> in the quantum state, where j ranges from 0 to 2n-1.

[0098] In order to simulate the dynamic changes of risk situation over time, a quantum risk evolution operator U is designed. el , (This operator is based on the quantum state of the current risk level, and deduces the risk situation at a certain point in the future under a given time step △t and environmental variable control parameter θ) The expression for simulating the change of risk situation space over time is: ; Where R is the total number of risk levels;

[0099] Preferably, a super position state |Ψ is formed sim >It contains information about the evolution of all risk levels over time and is a linear combination of all risk level states, thus allowing multiple risk scenarios to be considered simultaneously;

[0100] Based on the formation of a series of quantum states of super-position states, information is extracted from the super-position states through quantum measurement, and the effect expression of quantifying different risk response strategies is: ; Where E(O) is the expected value function, O is the observation operator of the physical quantity for evaluating the effectiveness of risk response strategies, and represents the effectiveness of a certain risk response strategy;

[0101] Combining the results of quantum evolution simulation and quantum measurement, by comparing the expected values ​​E(O) of different strategies, the risk response measure with the highest expected benefit is selected (which means that the system can automatically identify the most theoretically effective risk mitigation or response measure under the current risk situation), expressed as: ; Among them, sy ol is the optimal risk response strategy selected through the optimization process. It means finding the value of s that maximizes the expected value function among all risk response strategies;

[0102] It should be noted that the entire process utilizes the parallelism and efficiency of quantum computing, which greatly accelerates the process of risk simulation and response strategy selection, improves the response speed and accuracy of the driving safety warning system, and reflects the unique advantages of quantum computing in complex system risk management.

[0103] S5. Utilize edge computing technology to execute early warning decision-making algorithms on the edge data nodes of the Internet of Vehicles, and instantly generate early warning signals based on the optimal risk response measures. Through precise positioning via the Internet of Vehicles platform, the early warning signals are instantly transmitted to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow.

[0104] Furthermore, edge computing devices (these devices need to have strong data processing capabilities and low-latency communication capabilities for rapid processing and response) and efficient data processing and transmission middleware (used to receive data from multiple sources such as on-board sensors, roadside units (RSUs), cameras, etc. in real time to achieve preliminary data integration) are deployed on data nodes at the edge traffic fortresses of the Internet of Vehicles, city entrances, and accident-prone areas;

[0105] The data resources collected from edge computing nodes are optimized using an adaptive dynamic risk assessment model combined with a quantum enhancement algorithm to improve the sensitivity and accuracy of risk identification;

[0106] When the quantum-enhanced algorithm optimization analysis shows that the probability of a risk state exceeds the preset threshold, an early warning is immediately triggered and an early warning decision is deployed. This step uses the parallel processing capabilities of quantum computing to quickly complete the simulation and evaluation of a large number of possible risk scenarios;

[0107] It should be noted that "when the quantum enhanced algorithm optimization analysis shows that the probability of the risk state exceeds the preset threshold, the warning is immediately triggered and the warning decision is deployed", which specifically includes the following:

[0108] First, the pre-processed edge node data, including vehicle dynamic information, road conditions, weather conditions, etc., is input into the processing system of the quantum enhanced algorithm in real time. The quantum system uses the superposition and entanglement characteristics of quantum bits to perform high-speed parallel analysis of risk factors in complex environments;

[0109] During the operation of the algorithm, the probability of each potential risk occurring is quickly calculated by simulating the quantum state evolution of different risk factors. These probabilities reflect the current risk status of the vehicle or traffic network.

[0110] A risk tolerance threshold is set in advance based on historical data analysis and safety standards. This threshold represents the critical point for triggering an early warning. The system monitors the calculated risk probability in real time and compares it with the preset threshold. If the probability of any risk factor exceeds this threshold, the system immediately marks it as an abnormal state.

[0111] Based on the degree of exceeding the threshold and specific risk factors, the system automatically determines the urgency of the warning, such as minor warning, moderate warning or severe warning. Based on the warning level, the system automatically generates a warning signal containing key information such as risk type, urgency, and affected area. The warning information design should be concise and clear to ensure that the recipient can quickly understand and respond;

[0112] When the warning signal is triggered, the built-in warning decision algorithm is immediately activated. Based on the risk assessment results, the algorithm takes into account factors such as traffic flow, emergency resource distribution, and historical event handling experience to formulate the most appropriate response measures. The formulated warning decision is quickly distributed to relevant nodes through the Internet of Vehicles platform, including the nearest vehicle, traffic management center, and even the vehicle information system of a specific driver. At the same time, the corresponding infrastructure (such as smart traffic lights and roadside display screens) will also receive instructions and prepare to implement corresponding emergency measures, such as adjusting the timing of traffic lights and opening alternate routes.

[0113] After deploying the warning decision, the system continuously collects feedback information and evaluates the effectiveness of the warning measures, such as changes in traffic flow and accident avoidance. Based on the feedback results, the system may need to dynamically adjust the warning threshold, optimize the decision algorithm or adjust the response strategy to ensure that future warnings and response measures are more accurate and effective.

[0114] Furthermore, with the help of the vehicle's own perception data and multiple external information, it is transmitted in real time to the nearest edge computing node through the V2X communication network (including but not limited to V2V (vehicle-to-vehicle) and V2I (vehicle-to-infrastructure), to achieve real-time data transmission and ensure that the information can be quickly transmitted to the nearest edge computing node), and the received data is initially cleaned, including removing invalid data and duplicate data, and unifying the data format to ensure the efficiency and accuracy of subsequent processing;

[0115] Input the pre-processed data into the deployed early warning decision-making, and use the quantum-enhanced high-precision digital twin model to quickly determine the risk level of the vehicle's environment;

[0116] According to the risk level of the environment in which the vehicle is located, the edge node instantly extracts information from the super position state, calculates the expected benefits of different response strategies through quantum measurement, and selects the optimal risk response strategy with the highest expected benefit as a guide for action. This step uses the characteristics of quantum computing to accelerate the process of finding the best solution.

[0117] Based on the selected action guide (such as slowing down, changing routes or initiating emergency response procedures to effectively avoid risks), the edge node instantly generates visual warnings and sound alarms, and transmits the warning information to the vehicle closest to the risk area and the infrastructure node responsible for regulating traffic flow through the V2V communication network;

[0118] It should be noted that through this series of closely connected steps, the system realizes an efficient closed-loop process from data collection, processing, risk assessment to early warning decision-making, greatly improving the accuracy and response speed of driving safety warnings in the Internet of Vehicles environment, and effectively improving the level of road safety.

[0119] This embodiment also provides a driving safety warning system based on the Internet of Vehicles, including a perception and communication module, a data processing and fusion module, an intelligent analysis and decision-making module, an edge computing and warning module, and an integrated control and interaction module; the perception and communication module is used to collect vehicle surrounding environment data in real time, and communicate between vehicles and between vehicles and infrastructure through the vehicle-mounted Internet of Vehicles communication architecture to generate the vehicle's own perception data; the data processing and fusion module is used to optimize the timestamp matching of the vehicle surrounding environment data collected in real time, and construct a quantum-enhanced high-precision digital twin model; the intelligent analysis and decision-making module is used to use deep learning to drive the data fusion algorithm , the vehicle's own perception data is integrated with the multivariate external information received through the vehicle-mounted Internet of Vehicles for spatiotemporal multimodal data fusion to generate an adaptive dynamic risk assessment result; the edge computing and early warning module is used to simulate the risk scenario of the adaptive dynamic risk assessment result through a quantum-enhanced high-precision digital twin model to form the optimal risk response measures; the integrated control and interaction module is used to use edge computing technology to execute the early warning decision algorithm on the edge data node of the Internet of Vehicles, and generate early warning signals in real time according to the optimal risk response measures, and accurately locate them through the Internet of Vehicles platform and immediately transmit them to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow.

[0120] This embodiment also provides a computer device, which is suitable for the case of a driving safety warning method based on the Internet of Vehicles, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the driving safety warning method based on the Internet of Vehicles proposed in the above embodiment.

[0121] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covered on the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0122] The present embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, the driving safety warning method based on the Internet of Vehicles proposed in the above embodiment is implemented; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0123] In summary, the present invention aims to significantly improve the accuracy and response speed of driving safety warnings through a driving safety warning system and method based on the Internet of Vehicles by integrating quantum-enhanced high-precision digital twin models, deep learning-driven data fusion algorithms, and edge computing technology. It not only overcomes the shortcomings of the existing technology in real-time data processing, multimodal fusion, and risk assessment, but also through the introduction of quantum computing, it realizes efficient simulation of risk scenarios and selection of optimal response strategies, providing a solution with broad application value for solving the core problems of the current Internet of Vehicles warning system, such as untimely warnings and inaccurate risk assessments.

[0124] Example 2, referring to Table 1, is the second example of the present invention. To further verify the advancement of the present invention, experimental simulation data of a driving safety warning method based on the Internet of Vehicles is provided.

[0125] Four different test subjects were selected to simulate typical driving scenarios in the city: driving on a highway (scenario A), congested roads in the city (scenario B), driving at night on rainy days (scenario C), and detouring around a construction area (scenario D). Each vehicle was equipped with a sensor array as described in the invention and connected to an advanced Internet of Vehicles architecture.

[0126] First, in each scenario, the vehicle's sensor array (including high dynamic range night vision cameras, millimeter wave radars, lidars and ultrasonic sensors) works continuously, collecting at least 100 data samples per second to ensure the real-time and accuracy of environmental perception. The data includes but is not limited to key parameters such as vehicle distance, vehicle speed, object recognition, and road conditions.

[0127] Secondly, the original data was effectively denoised and optimized through Kalman filtering and image processing algorithms. Subsequently, the quantum enhancement algorithm was used to optimize timestamp matching, which improved the accuracy of data synchronization. With the support of the quantum digital twin model, these data were further converted into a highly accurate environmental model.

[0128] Next, combined with a deep learning-driven data fusion algorithm, the vehicle’s own perception data and external information (such as traffic signals and weather conditions) are fused in a spatiotemporal multimodal manner to generate an adaptive dynamic risk assessment result. This step utilizes a deep neural network model to achieve accurate classification of different risk levels.

[0129] Finally, based on the quantum-enhanced high-precision digital twin model, risk scenario simulations were performed to quickly determine the optimal risk response measures for each scenario. The edge computing device then generates early warning signals in real time based on these measures.

[0130] The details are shown in the following table:

[0131] Table 1 Experimental record table

[0132] Comparison of table data shows that the driving safety warning method of the present invention is significantly superior to the traditional warning system in multiple key indicators. For example, in scenario C (driving at night on rainy days), the vehicle distance error is only 0.03 meters, which is 75% lower than the 0.12 meters of the traditional system. This directly reflects the advantage of the quantum enhanced digital twin model in improving the accuracy of environmental perception; at the same time, the risk response time is reduced by 60% on average, and the minimum can be as low as 0.1 second, which greatly improves the response efficiency in emergency situations; the warning accuracy is increased to more than 96%, which is significantly higher than the 85% of the traditional system, indicating that the deep learning fusion algorithm has obvious advantages in multi-source data processing; the data processing delay is reduced by 75% on average, which ensures the real-time nature of information processing and helps to make decisions faster; in addition, the improvement in risk assessment efficiency ensures the timeliness and effectiveness of the warning signal, significantly enhancing road safety.

[0133] The present invention significantly enhances the accuracy and efficiency of driving safety warnings by integrating cutting-edge technologies such as vehicle networking communications, quantum information technology, deep learning, and edge computing. Specifically, it realizes real-time, high-precision collection and processing of vehicle surrounding environment data, and uses quantum-enhanced high-precision digital twin models and deep learning-driven data fusion algorithms to greatly reduce vehicle distance errors, shorten risk response time, and increase warning accuracy to more than 98%. It also reduces data processing delays and ensures the timeliness of information processing. In addition, the present invention can adaptively generate optimal risk response strategies, accurately locate and instantly transmit warning signals through the vehicle networking platform, effectively cover vehicles and transportation infrastructure closest to risk areas, and comprehensively improve the safety management level and emergency response capabilities of road traffic, bringing revolutionary innovation and progress to the field of safety warnings for intelligent transportation systems.

[0134] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A driving safety warning method based on Internet of Vehicles, characterized by: include, Collect data about the vehicle's surrounding environment in real time, communicate between vehicles and between vehicles and infrastructure through the vehicle-to-vehicle networking communication architecture, and generate vehicle-sensing data; Perform timestamp matching optimization on the real-time collected vehicle surrounding environment data to build a quantum-enhanced high-precision digital twin model; Using deep learning driven data fusion algorithms, the vehicle’s own perception data is integrated with the multivariate external information received through the vehicle-mounted Internet of Vehicles to generate adaptive dynamic risk assessment results; Through the quantum-enhanced high-precision digital twin model, the adaptive dynamic risk assessment results are simulated for risk scenarios to form the optimal risk response measures; Using edge computing technology, early warning decision algorithms are executed on the edge data nodes of the Internet of Vehicles, and early warning signals are generated instantly based on the optimal risk response measures. Through the Internet of Vehicles platform, accurate positioning is performed and the early warning signals are instantly transmitted to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow. The specific steps of performing timestamp matching optimization on the real-time collected vehicle surrounding environment data and building a quantum enhanced high-precision digital twin model are as follows: Based on the driving sensor data, quantum entanglement is used to perform preprocessing and timestamp matching optimization. The expression is: ; in, is a quantum entangled state, and represent the states of the first and second qubits respectively; Based on the exchange mechanism in the quantum entangled state, the timestamps are matched and the synchronized data is represented as a quantum state. ; Using quantum feature mapping, the synchronized quantum state Mapped to the feature space, the quantum state expression after feature extraction is: ; in, is the quantum state after feature extraction, U ft Quantum operations for feature mapping; The expression for training a set of quantum states of the support vector machine based on the quantum state after feature extraction is: ; in, is the quantum state of the i-th training sample, and M is the number of samples in the training set; Based on the trained quantum state, the similarity expression is calculated by quantum inner product: ; in, For test state and the quantum state of the i-th training sample The similarity between is the feature mapping quantum operation U ft The transposed conjugate of Substitute the similarity value calculated by quantum inner product into the decision rule of quantum support vector machine, and use quantum version of Lagrange multiplier method and kernel technique to make prediction expression of quantum support vector machine: ; in, represents the classification decision result of the test state based on the quantum support vector machine, sign is the sign function that determines the positive or negative of the classification result, α i is the Lagrange multiplier of the i-th training sample, y i is the label of the i-th training sample, and b is the bias term for adjusting the position of the classification hyperplane; The classification decision result of the test state based on the quantum support vector machine is used as a classification indication of the current vehicle environment state and integrated into the dynamic update mechanism of the quantum digital twin model to construct a quantum enhanced high-precision digital twin model. The expression is: ; Where t represents the time step, is the quantum state of the vehicle environment at time t, U ee is the evolution operator that directly responds to the quantum support vector classification result, and U is the operator; The deep learning driven data fusion algorithm is used to perform spatiotemporal multimodal data fusion of the vehicle's own perception data and the multivariate external information received through the vehicle-mounted Internet of Vehicles to generate an adaptive dynamic risk assessment result. The specific steps are as follows: Receive multiple external information such as traffic signal status, weather forecast, road construction information, vehicle location and speed through the vehicle-mounted Internet of Vehicles; Based on the vehicle's own perception data and the multivariate external information received by the vehicle-mounted Internet of Vehicles, the GPS positioning data and timestamp matching method are combined for time synchronization and spatial registration to form fused data; For the image data and radar data in the vehicle's own perception data, deep convolutional neural networks and point cloud processing algorithms are used to extract high-level semantics and multi-modal features of the object's motion trajectory; The spatiotemporal attention mechanism STAM is used to fuse the extracted multimodal features: ; Among them, F fn is the fused feature vector, F ie , F rr , ..., F el Represent the feature vectors extracted from image, radar and external information respectively; Based on the fused feature vector, a deep neural network model is constructed; The input layer of the deep neural network model accepts the fused feature vector expression as: ; ; Among them, a (l) is the activation output of the lth layer, f (l) is the ReLU activation function used in the lth layer, z (l) is the weighted input of the lth layer, W (l) is the weight matrix from layer l−1 to layer l, b (l) is the bias vector of the lth layer; For the output layer of the deep neural network model, the risk level expression is obtained through the softmax function: ; Among them, P(y=k|F fn ) is the feature vector F after fusion fn Under the condition of , the probability that the true risk level label y belongs to the category label k, e zk is the linear weighted input value z of the output node for category k, C is the total number of categories, and j is the category index; The cross entropy loss function is selected as the target of supervised learning of the deep neural network model, and the Adam optimizer is used to update the model parameters. The loss function model L(y, ) is expressed as: ; Among them, y o,c is the value of the true risk level label y under sample o and category c, The probability distribution predicted by the loss function model; After the cross entropy loss function is performed, the optimization fitting of the deep neural network model is completed, and the adaptive dynamic risk assessment results are generated by receiving the real-time input of the vehicle's own perception data and the multivariate external information received by the vehicle-mounted Internet of Vehicles; The quantum-enhanced high-precision digital twin model is used to simulate risk scenarios for adaptive dynamic risk assessment results to form optimal risk response measures. The specific steps are as follows: Each risk level in the adaptive dynamic risk assessment result is mapped to a quantum state, and the quantum risk scenario space expression is constructed by converting binary into quantum bit sequence: ; in, represents the state of the i-th risk level, is the quantum ground state, n represents the quantum resource amount encoding risk level information, a is the amplitude coefficient, i and j are the subscripts of risk level and quantum ground state respectively; Design of quantum risk evolution operator U el , the expression of the simulated risk situation space changing over time is: ; Among them, U el is the quantum operator that controls the evolution of risk situation, △t is the time step, and θ is the environmental variable control parameter; Using the parallel processing capabilities of quantum computers, quantum evolution is used to simulate the vehicle behavior response under various risk scenarios to form a super position state. The expression is: ; Among them, R is the total number of risk levels; Based on the formation of a series of quantum states of super-position states, information is extracted from the super-position states through quantum measurement, and the effect expression of quantifying different risk response strategies is: ; Among them, E(O) is the expected value function, and O is the observation operator of the physical quantity that evaluates the effectiveness of the risk response strategy; Combining the quantum evolution simulation and quantum measurement results, the risk response measures with the highest expected benefits are selected, expressed as: ; Among them, sy ol is the optimal risk response strategy selected through the optimization process. It refers to finding the value s at which the expected value function reaches the maximum value among all risk response strategies.

2. The driving safety warning method based on the Internet of Vehicles according to claim 1, characterized in that: The real-time collection of vehicle surrounding environment data, communication between vehicles and between vehicles and infrastructure through the vehicle-to-vehicle network communication architecture, and generation of vehicle-sensing data are specifically performed as follows: Various sensors such as high dynamic range night vision cameras, millimeter wave radars, lidars, and ultrasonic sensors are installed at key locations on the front and rear bumpers, under the side mirrors, and on the roof of the vehicle; The various sensors installed collect data about the vehicle's surrounding environment, such as images, distance, and speed, and use a timestamp synchronization mechanism to output initial sensor data; The Kalman filter algorithm is used to remove noise from the initial sensor data, and the background subtraction and edge detection image processing algorithms are used to optimize the quality of camera image data. Cluster analysis is performed through radar point clouds to distinguish static and dynamic objects, and integrated to form standard driving sensor data; Aggregate standard driving sensor data from the vehicle bus to a central processor for image frames, radar point clouds, ultrasonic echoes and vehicle posture information; The camera images and radar point cloud data from standard driving sensor data are clustered and analyzed, combined with machine vision and YOLOv5 deep learning algorithm to build a 3D environment model. The integrated communication protocol supporting V2V and V2I will be connected to the Internet of Vehicles, integrating the three-dimensional environmental model and standard driving sensor data, and communicating between vehicles and between vehicles and infrastructure through the on-board communication architecture to generate the vehicle's own perception data.

3. The driving safety warning method based on the Internet of Vehicles as claimed in claim 2, characterized in that: The edge computing technology is used to execute early warning decisions on the edge data nodes of the Internet of Vehicles, and early warning signals are generated instantly according to the optimal risk response measures. The specific steps are as follows: Deploy edge computing devices and efficient data processing and transmission middleware at data nodes at edge traffic fortresses, city entrances, and accident-prone areas of the Internet of Vehicles; The collected data resources of edge computing nodes are optimized using an adaptive dynamic risk assessment model combined with a quantum enhancement algorithm; When the quantum enhanced algorithm optimization analysis shows that the probability of the risk state exceeds the preset threshold, an early warning is immediately triggered and an early warning decision is deployed.

4. The driving safety warning method based on the Internet of Vehicles as claimed in claim 3, characterized in that:The precise positioning through the Internet of Vehicles platform is immediately transmitted to the vehicle closest to the risk area and the infrastructure node responsible for regulating traffic flow. The specific steps are as follows: With the help of the vehicle's own perception data and multiple external information, it is transmitted to the nearest edge computing node in real time through the V2X communication network, and the received data is initially cleaned, deduplicated and formatted; Input the pre-processed data into the deployed early warning decision-making, and use the quantum-enhanced high-precision digital twin model to quickly determine the risk level of the vehicle's environment; According to the risk level of the environment in which the vehicle is located, the edge node instantly extracts information from the super position state, calculates the expected benefits of different response strategies through quantum measurement, and selects the optimal risk response strategy with the highest expected benefit as a guide for action; Based on the selected action guidelines, the edge node instantly generates visual warnings, sound alarms, and transmits the warning information to the vehicles closest to the risk area and the infrastructure nodes responsible for regulating traffic flow through the V2V communication network.

5. A vehicle networking-based driving safety warning system, based on the vehicle networking-based driving safety warning method according to any one of claims 1 to 4, characterized in that: Including perception and communication module, data processing and fusion module, intelligent analysis and decision-making module, edge computing and early warning module, integrated control and interaction module; The perception and communication module is used to collect vehicle surrounding environment data in real time, communicate between vehicles and between vehicles and infrastructure through the vehicle-to-vehicle network communication architecture, and generate vehicle-sensing data; The data processing and fusion module is used to optimize the timestamp matching of the vehicle surrounding environment data collected in real time and build a quantum-enhanced high-precision digital twin model; The intelligent analysis and decision-making module is used to use a deep learning-driven data fusion algorithm to perform spatiotemporal multimodal data fusion of the vehicle's own perception data and the multivariate external information received through the vehicle-mounted Internet of Vehicles to generate an adaptive dynamic risk assessment result; The edge computing and early warning module is used to simulate risk scenarios of adaptive dynamic risk assessment results through quantum-enhanced high-precision digital twin models to form optimal risk response measures; The integrated control and interaction module is used to utilize edge computing technology to execute an early warning decision algorithm on the edge data nodes of the Internet of Vehicles, generate an early warning signal in real time based on the optimal risk response measures, and accurately locate the early warning signal through the Internet of Vehicles platform and transmit it in real time to the vehicle closest to the risk area and the infrastructure node responsible for regulating traffic flow.

6. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the driving safety warning method based on the Internet of Vehicles as described in any one of claims 1 to 4 are implemented.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the driving safety warning method based on the Internet of Vehicles as described in any one of claims 1 to 4 are implemented.