Holographic road network and Internet of Vehicles fusion method and device, equipment and storage medium

Abstract

The invention relates to the technical field of Internet of Vehicles, and discloses a holographic road network and Internet of Vehicles fusion method and device, equipment and a storage medium. The method comprises the steps of obtaining multi-source heterogeneous data collected by a target sensor, and performing adaptive multi-modal preprocessing on the multi-source heterogeneous data; performing space-time alignment and hierarchical structure recombination on the preprocessed multi-source heterogeneous data to obtain multi-source traffic recombination data; extracting multi-scale spatio-temporal features in the multi-source traffic recombination data, and performing multi-level semantic association, deep probability fusion and uncertainty modeling on the multi-source traffic recombination data based on the extracted traffic feature vector set and the preliminary dynamic traffic knowledge representation to obtain traffic perception probability distribution; and a high-precision dynamic semantic map is constructed based on traffic perception probability distribution, and fused multi-resolution map representation is obtained. According to the invention, the fusion of the holographic road network and the Internet of Vehicles multi-source data is realized, and the accuracy of the traffic characteristic data in the fusion result is improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicle networking technology, and in particular to a method, device, equipment and storage medium for integrating a holographic road network with a vehicle networking. Background Art

[0002] In the field of intelligent transportation systems, the integration of holographic road networks and vehicle networks is a key technology for achieving efficient, safe and sustainable traffic management. The holographic road network provides accurate road infrastructure information, while the vehicle network captures vehicle dynamic data in real time. Any effective integration of these two systems is of great significance for optimizing traffic flow, improving road utilization and enhancing traffic safety.

[0003] Nowadays, attempts are made to integrate holographic road network and IoV data using data fusion algorithms and machine learning techniques, or to apply edge computing methods to process large amounts of real-time information. However, these methods still face challenges in data heterogeneity processing, real-time assurance, and dynamic environment adaptation. In addition, existing fusion methods often ignore some unique characteristics of the transportation system, such as the complexity of the road network topology, the randomness of vehicle behavior, and the spatiotemporal changes in traffic flow patterns, which may have a significant impact on the fusion effect. That is, the existing fusion methods of holographic road networks and IoVs fail to fully consider the characteristics of multi-source data and the unique properties of the transportation system, resulting in low accuracy of the fusion results. Summary of the invention

[0004] The main purpose of the present invention is to solve the problem that the fusion method of holographic road network and vehicle network fails to fully consider the characteristics of multi-source data and the unique properties of the transportation system, resulting in low accuracy of the fusion result.

[0005] The first aspect of the present invention provides a method for integrating a holographic road network and an Internet of Vehicles, the method comprising: acquiring multi-source heterogeneous data of the entire road network and the Internet of Vehicles collected by a target sensor, and performing adaptive multimodal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; performing spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and performing hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganization data; extracting multi-scale spatiotemporal features from the multi-source traffic reorganization data to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, performing multi-level semantic association on the multi-source traffic reorganization data to obtain an enhanced dynamic knowledge representation; based on the enhanced dynamic knowledge representation, performing deep probability fusion and uncertainty modeling on the multi-source traffic reorganization data to obtain a traffic perception probability distribution, and constructing a high-precision dynamic semantic map based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

[0006] Optionally, in a first implementation method of the first aspect of the present invention, the preprocessed multi-source heterogeneous data is subjected to spatiotemporal alignment to obtain spatiotemporally aligned multi-source heterogeneous data, including: performing time series analysis and frequency domain decomposition on the multi-source heterogeneous data to obtain multi-frequency band time components, and performing window calculation of each frequency band and time series merging and sorting on the multi-frequency band time components to obtain multi-resolution multi-source heterogeneous time series data; performing spatial alignment of multi-target sensors and multi-source clock synchronization on the multi-source heterogeneous time series data to obtain spatiotemporal reference multi-source data, and performing dynamic coordinate system conversion on the spatiotemporal reference multi-source data to obtain global coordinate multi-source data of a global reference system; performing data stream alignment of multi-target sensors on the global coordinate multi-source data to obtain a synchronized multi-source data stream, and constructing a multi-dimensional spatiotemporal index corresponding to the synchronized multi-source data stream to obtain spatiotemporally aligned multi-source heterogeneous data.

[0007] Optionally, in a second implementation method of the first aspect of the present invention, the hierarchical structural reorganization of the spatiotemporally aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data includes: performing multimodal anomaly detection on the spatiotemporally aligned multi-source heterogeneous data to obtain multi-source anomaly labeled data, and performing context-aware data repair on the multi-source anomaly labeled data to obtain preliminarily repaired multi-source heterogeneous data; performing adaptive multi-scale filtering on the preliminarily repaired multi-source heterogeneous data to obtain noise-suppressed multi-source heterogeneous data, and performing multimodal data traffic scene segmentation and fusion on the noise-suppressed multi-source heterogeneous data to obtain a semantically segmented multi-source traffic scene; performing traffic key element extraction and traffic attribute labeling on the semantically segmented multi-source traffic scene to obtain a multi-source traffic structured representation, and performing spatiotemporal dynamic characteristic modeling on the multi-source traffic structured representation to obtain multi-source traffic reorganized data.

[0008] Optionally, in a third implementation method of the first aspect of the present invention, the extracting of multi-scale spatiotemporal features from the multi-source traffic reorganization data to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation includes: extracting road network topology features and vehicle trajectory features from the multi-source traffic reorganization data to obtain an initial road network and vehicle feature set, and performing time series segmentation and periodic analysis on the initial road network and vehicle feature set to obtain multi-scale traffic flow features; performing segment-level and regional-level spatial aggregation on the multi-scale traffic flow features to obtain a multi-level traffic state representation, and performing a road network topology-guided long-range dependency analysis on the multi-level traffic state representation to obtain global traffic correlation features; based on preset traffic scene events, performing dynamic feature screening and fusion on the global traffic correlation features to obtain a traffic scene representation, and performing traffic knowledge graph construction and rule reasoning on the traffic scene representation to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, the multi-source traffic reorganization data is subjected to multi-level semantic association to obtain an enhanced dynamic knowledge representation, including: performing multi-scale matching of the road network structure and the vehicle trajectory on the traffic feature vector set to obtain a road network and vehicle association mapping, and performing spatiotemporal consistency constraints and dynamic smoothing processing on the road network and vehicle association mapping to obtain traffic element features with fine feature alignment; constructing a hierarchical traffic scene graph for the preliminary dynamic traffic knowledge representation of the traffic element features to obtain a multi-level traffic semantic network, and based on preset traffic driving specifications, performing contextual reasoning and correlation analysis of traffic rules on the multi-level traffic semantic network to obtain a traffic semantic association strength matrix; based on preset historical traffic patterns, performing iterative reasoning and semantic description detection on the traffic semantic association strength matrix to generate a traffic scene semantic relationship set, and performing traffic anomaly detection and relationship dynamic updating on the traffic scene semantic relationship set to obtain a time-varying traffic knowledge graph; performing multi-source information fusion and traffic driving knowledge reasoning on the time-varying traffic knowledge graph and the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, based on the enhanced dynamic knowledge representation, the multi-source traffic reorganization data is subjected to deep probability fusion and uncertainty modeling to obtain a traffic perception probability distribution, including: constructing a multi-level traffic network probability graph for the enhanced dynamic knowledge representation to obtain a hierarchical traffic state probability distribution, and based on preset historical traffic data, performing a spatiotemporal consistency comparison on the hierarchical traffic state probability distribution and adjusting the weights of the corresponding preliminary dynamic traffic knowledge representations at each level to obtain data confidence weights of the corresponding preliminary dynamic traffic knowledge representations at each level; performing Bayesian fusion on the data confidence weights and the hierarchical traffic state probability distribution to obtain an updated traffic state posterior probability, and performing spatiotemporal correlation analysis and Markov random field modeling on the traffic state posterior probability to generate a dynamic traffic system evolution model; performing multi-objective traffic optimization solution on the dynamic traffic system evolution model to obtain a traffic state prediction and a traffic control strategy set, and performing real-time traffic data verification and updating on the traffic state prediction and the traffic control strategy set to obtain a traffic perception probability distribution.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, a high-precision dynamic semantic map is constructed based on the traffic perception probability distribution to obtain a fused multi-resolution map representation, including: extracting a multi-level traffic network topology structure from the traffic perception probability distribution to obtain a preliminary traffic semantic skeleton diagram, and performing spatiotemporal interpolation and dynamic feature completion on the preliminary traffic semantic skeleton diagram to obtain a continuous multi-dimensional traffic scene representation; based on the traffic perception probability distribution, performing probability evaluation and identifying uncertainty areas on the continuous multi-dimensional traffic scene representation to obtain a region recognition result, and based on the time-varying traffic knowledge graph , perform semantic supplementation on the uncertainty area to obtain a semantic supplementation result; based on the semantic supplementation result, perform consistency constraints and map integration on the continuous multi-dimensional traffic scene representation to generate a dynamic traffic semantic map, and perform traffic flow change detection and local update on the dynamic traffic semantic map to generate a time-varying traffic network model; perform multi-scale traffic feature extraction and representation scheme optimization on the time-varying traffic network model to obtain an adaptive multi-resolution traffic map, and perform spatiotemporal index construction and evolution analysis on the adaptive multi-resolution traffic map to obtain a final multi-resolution map representation of the fusion of the holographic road network and the Internet of Vehicles.

[0012] The second aspect of the present invention provides a holographic road network and vehicle network fusion device, which includes: a preprocessing module for acquiring multi-source heterogeneous data of the entire road network and vehicle network collected by the target sensor, and performing adaptive multi-modal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; a structural reorganization module for performing spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and performing hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data; a semantic association module for The multi-scale spatiotemporal features in the multi-source traffic reorganization data are extracted to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, the multi-source traffic reorganization data is subjected to multi-level semantic association to obtain an enhanced dynamic knowledge representation; a map representation module is used to perform deep probability fusion and uncertainty modeling on the multi-source traffic reorganization data based on the enhanced dynamic knowledge representation to obtain a traffic perception probability distribution, and a high-precision dynamic semantic map is constructed based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

[0013] The third aspect of the present invention provides a holographic road network and vehicle network fusion device, comprising: a memory and at least one processor, wherein the memory stores instructions; the at least one processor calls the instructions in the memory to enable the holographic road network and vehicle network fusion device to perform each step of the above-mentioned holographic road network and vehicle network fusion method.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium, which stores instructions. When the computer-readable storage medium is run on a computer, it enables the computer to execute each step of the above-mentioned method for integrating holographic road network and vehicle network.

[0015] The above-mentioned holographic road network and vehicle network integration method, device, equipment and storage medium. In the embodiment of the present invention, the multi-source heterogeneous data of the entire road network and vehicle network are collected by acquiring target sensors, and the multi-source heterogeneous data are adaptively multimodally preprocessed to obtain the preprocessed multi-source heterogeneous data; the preprocessed multi-source heterogeneous data are time-space aligned to obtain time-space aligned multi-source heterogeneous data, and the time-space aligned multi-source heterogeneous data are hierarchically reorganized to obtain multi-source traffic reorganized data; the multi-scale time-space features in the multi-source traffic reorganized data are extracted to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, the multi-source traffic reorganized data is multi-level semantically associated to obtain an enhanced dynamic knowledge representation; based on the enhanced dynamic knowledge representation, the multi-source traffic reorganized data is deeply probabilistically fused and uncertainty modeled to obtain a traffic perception probability distribution, and a high-precision dynamic semantic map is constructed based on the traffic perception probability distribution to obtain a fused multi-resolution map representation. Compared with the prior art, the present application performs adaptive multimodal preprocessing and spatiotemporal alignment on the acquired multi-source heterogeneous data, and then hierarchically restructures the spatiotemporal aligned data, extracts multi-scale spatiotemporal features, constructs a preliminary dynamic traffic knowledge representation, and then performs multi-level semantic association on the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation. Based on the enhanced dynamic knowledge representation, deep probability fusion and uncertainty modeling are performed on the multi-source traffic reorganized data to obtain the traffic perception probability distribution, and then a high-precision dynamic semantic map is constructed to form a fused multi-resolution map representation. The present application enhances the ability to understand and express complex traffic scenes through multi-level processing and semantic association, and integrates probability models and uncertainty analysis, thereby improving the system's adaptability and prediction capabilities for dynamic traffic environments, realizing the effective fusion of holographic road network and Internet of Vehicles data, and improving the comprehensiveness and accuracy of traffic feature data in the fusion results.

[0016] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0017] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is a schematic diagram of a first embodiment of a method for integrating a holographic road network with an Internet of Vehicles in an embodiment of the present invention; Figure 2 It is a schematic diagram of an embodiment of a holographic road network and vehicle network integration device in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of a holographic road network and vehicle network integration device in an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0020] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0021] To facilitate understanding of this embodiment, the specific process of the embodiment of the present invention is described below. Figure 1 The first embodiment of the method for integrating holographic road network and vehicle network in the embodiment of the present invention includes: 101. Obtain multi-source heterogeneous data collected by target sensors from the entire road network and vehicle network, and perform adaptive multi-modal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Artificial Intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0022] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0023] In this embodiment, the above-mentioned multi-source heterogeneous data include but are not limited to the following categories: holographic road network data (such as high-precision map data, traffic infrastructure data and road condition information), Internet of Vehicles data (vehicle real-time location data, vehicle dynamic information, vehicle-mounted sensor data), traffic management system data, environmental data and commercial activity data, etc.

[0024] In practical applications, comprehensive road network data is acquired by deploying a self-organizing multimodal sensor network in advance (this network includes phased array lidar, multispectral camera array and millimeter wave radar grid, and adopts a dynamic layered sampling strategy, which can adaptively adjust the sampling frequency and accuracy according to the complexity of the environment and the importance of the data, thereby ensuring the collection of high-quality holographic road network data). At the same time, in order to obtain Internet of Vehicles data, intelligent edge computing units are installed on the vehicle. These units can not only collect high-precision GPS location information, but also integrate inertial measurement units to provide multi-dimensional acceleration and attitude information, which greatly improves the accuracy and reliability of vehicle dynamic data. In addition, while collecting data, an adaptive multi-level data filtering mechanism is implemented at the edge computing node, that is, first, a reconfigurable hardware-level filtering based on FPGA is used to quickly remove obvious noise and outliers at the data source; second, an adaptive extended Kalman filter is used to dynamically adjust the filtering parameters according to the data characteristics. Effectively smooth time series data; finally, semantic-level filtering based on the graph attention network understands the contextual relationship of the data, identifies and processes anomalies at the semantic level, thereby greatly improving the quality and availability of the data through a multi-level filtering mechanism; then, a distributed spatiotemporal alignment algorithm is used to combine dynamic time warping technology and spatial transformation network to accurately align data from different sources while processing large-scale data streams in real time. Dynamic time warping technology is used to deal with problems of different sampling rates and time delays, while the spatial transformation network learns complex spatial transformation relationships, thereby unifying data in different coordinate systems into a consistent representation; then, an adaptive spatial transformation network is used to learn the mapping relationship between different coordinate systems in the data after spatial changes, and dynamically adjusts the conversion parameters according to the characteristics of the data, thereby unifying the spatial data from different sensors and different vehicles into a dynamic global reference system, thereby obtaining multi-source heterogeneous data after adaptive multimodal preprocessing.

[0025] 102. Performing spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and performing hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data; In this embodiment, the multi-source heterogeneous data are subjected to time series analysis and frequency domain decomposition to obtain multi-frequency band time components, and the multi-frequency band time components are subjected to window calculation of each frequency band and time series merging and sorting to obtain multi-resolution multi-source heterogeneous time series data; the multi-source heterogeneous time series data are subjected to spatial alignment of multi-target sensors and multi-source clock synchronization to obtain spatiotemporal reference multi-source data, and the spatiotemporal reference multi-source data are subjected to dynamic coordinate system conversion to obtain global coordinate multi-source data of a global reference system; the global coordinate multi-source data are subjected to data stream alignment of multi-target sensors to obtain a synchronous multi-source data stream, and a multi-dimensional spatiotemporal index corresponding to the synchronous multi-source data stream is constructed to obtain a spatiotemporally aligned multi-source heterogeneous data. The method comprises the steps of: performing multimodal anomaly detection on the spatiotemporally aligned multi-source heterogeneous data to obtain multi-source anomaly labeled data, and performing context-aware data repair on the multi-source anomaly labeled data to obtain preliminarily repaired multi-source heterogeneous data; performing adaptive multi-scale filtering on the preliminarily repaired multi-source heterogeneous data to obtain noise-suppressed multi-source heterogeneous data, and performing multimodal data traffic scene segmentation and fusion on the noise-suppressed multi-source heterogeneous data to obtain semantically segmented multi-source traffic scenes; performing traffic key element extraction and traffic attribute labeling on the semantically segmented multi-source traffic scenes to obtain multi-source traffic structured representation, and performing spatiotemporal dynamic characteristic modeling on the multi-source traffic structured representation to obtain multi-source traffic reorganized data.

[0026] In practical applications, firstly, the time series analysis and frequency domain decomposition of multi-source heterogeneous data are performed by using continuous wavelet transform to decompose the time series data corresponding to the multi-source heterogeneous data into time components of multiple frequency bands (where these components represent the changing characteristics of the data at different time scales), and the appropriate window function is applied to each frequency band of these multi-frequency band time components for smoothing to reduce spectrum leakage and improve frequency resolution. At the same time, considering the characteristics of different frequency bands, an adaptive window size is adopted to recombine these processed time components, and by merging and sorting the time series, multi-source heterogeneous time series data with multiple resolutions is obtained (where these data retain the original information). At the same time, it also provides multi-scale time feature representation, where the multi-resolution can be hourly, daily and weekly multi-resolution representation); then the Kalman filter is extended to perform spatial registration of multi-target sensors for multi-source heterogeneous time series data, and the network time protocol (NTP) method is used to ensure the time consistency between different data sources to obtain spatiotemporal reference multi-source data (for example, the data of vehicle-mounted GPS, inertial measurement unit (IMU) and roadside radar are fused to obtain more accurate vehicle position and time information), and the quaternion interpolation algorithm is used to perform dynamic coordinate system conversion on the spatiotemporal reference multi-source data, unifying all data into a global reference system, where the quaternion interpolation formula is: ; Among them, q(t) is the interpolated quaternion, q0 and q1 are the starting and ending quaternions, Ω is the angle between the quaternions, and t is the interpolation parameter. The data in different coordinate systems are smoothly converted to the global reference system through the quaternion interpolation algorithm to obtain global coordinate multi-source data in the global reference system (for example, the vehicle-mounted sensor data and roadside equipment data are converted to a unified geographic coordinate system); then the dynamic time warping (DTW) algorithm is used to align the data streams of multi-target sensors on the global coordinate multi-source data to accurately align all data in the time dimension to obtain a synchronized multi-source data stream, and an R-tree index structure is used to construct a multidimensional spatiotemporal index corresponding to the synchronized multi-source data stream to obtain spatiotemporally aligned multi-source heterogeneous data (for example, by constructing an index structure, all vehicle trajectory data in a certain area within a specific time period can be quickly located).

[0027] Secondly, a deep autoencoder network is combined with an isolation forest algorithm to perform multimodal anomaly detection on the spatiotemporally aligned multi-source heterogeneous data. That is, the deep autoencoder captures normal patterns by learning a low-dimensional representation of the data, and the isolation forest is used to calculate the isolation of data points to identify anomalies, so as to identify various types of anomalies, such as sensor failure, communication interruption or abnormal traffic incidents, and obtain multi-source anomaly labeled data. A graph convolutional neural network (GCN) is used to model the spatial and temporal dependencies between the multi-source anomaly labeled data, and an attention mechanism is used to capture long-distance contextual information, so as to use the information of the surrounding normal data to repair the abnormal data, maintain the spatial and temporal consistency of the data, and obtain the preliminarily repaired multi-source heterogeneous data. Then, a wavelet packet transform (WPT) combined with an adaptive threshold technology is used to perform the preliminarily repaired data. Adaptive multi-scale filtering is performed on the multi-source heterogeneous data to further suppress noise, that is, WPT is used to perform more detailed frequency division on the preliminarily repaired multi-source heterogeneous data to accurately locate and process noise of different scales, and an adaptive threshold is used to dynamically adjust the threshold according to the statistical characteristics of each sub-band to ensure that useful information is retained to the greatest extent while denoising, so as to obtain noise-suppressed multi-source heterogeneous data, and a method combining a fully convolutional network (FCN) and a conditional random field (CRF) is used to segment and fuse the multi-modal data traffic scene of the noise-suppressed multi-source heterogeneous data, and FCN is used to process inputs of different modes (such as images, point clouds, radar data) to generate preliminary semantic segmentation results, and CRF is used to further optimize the segmentation boundary and improve the segmentation accuracy. The combination of the two can be expressed as: ; in, is the unary potential energy generated by FCN, is a pairwise potential, used to model the relationship between adjacent pixels or data points, is the set of nodes in the graph, is the edge set in the graph, (i, j) represents the edge between node i and node j, is the state or variable at node i, is the state or variable at node j. By minimizing this energy function, an accurate semantic segmentation result, i.e., a semantically segmented multi-source traffic scene, is obtained, and key traffic elements such as roads, vehicles, and pedestrians are identified. Then, the semantically segmented multi-source traffic scene is subjected to traffic key element extraction and traffic attribute labeling to accurately locate and classify traffic key elements. At the same time, the graph attention network (GAT) is used to model the relationship between elements and extract higher-level semantic information to obtain rich traffic attribute information, such as vehicle type, driving direction, meaning of traffic signs, etc. A combination of long short-term memory network (LSTM) and spatiotemporal graph convolutional network (ST-GCN) is used to model the spatiotemporal dynamic characteristics of the multi-source traffic structured representation to obtain multi-source traffic reorganization data, i.e., LSTM is used to capture long-term temporal dependencies. At the same time, ST-GCN is used to model spatial and temporal relationships to obtain a comprehensive and dynamic traffic network representation, which contains rich spatiotemporal information and semantic content.

[0028] 103. Extract multi-scale spatiotemporal features from multi-source traffic reorganization data to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, perform multi-level semantic association on the multi-source traffic reorganization data to obtain an enhanced dynamic knowledge representation; In this embodiment, the road network topology features and vehicle trajectory features of the multi-source traffic reorganization data are extracted to obtain an initial road network and vehicle feature set, and the initial road network and vehicle feature set are subjected to time series segmentation and periodic analysis to obtain multi-scale traffic flow features; the multi-scale traffic flow features are subjected to segment-level and regional-level spatial aggregation to obtain a multi-level traffic state representation, and the multi-level traffic state representation is subjected to a road network topology-guided long-range dependency analysis to obtain a global traffic association feature; based on preset traffic scene events, the global traffic association features are subjected to dynamic feature screening and fusion to obtain a traffic scene representation, and the traffic scene representation is subjected to traffic knowledge graph construction and rule reasoning to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation; the traffic feature vector set is subjected to multi-scale matching of the road network structure and the vehicle trajectory to obtain a road network and vehicle association map. The method comprises the following steps: mapping the road network and the vehicle association mapping, and performing spatiotemporal consistency constraints and dynamic smoothing processing on the road network and the vehicle association mapping to obtain traffic element features with fine feature alignment; constructing a hierarchical traffic scene graph for the preliminary dynamic traffic knowledge representation of the traffic element features to obtain a multi-level traffic semantic network, and based on the preset traffic driving specifications, performing contextual reasoning and correlation analysis on the multi-level traffic semantic network to obtain a traffic semantic association strength matrix; based on the preset historical traffic pattern, performing iterative reasoning and semantic description detection on the traffic semantic association strength matrix to generate a traffic scene semantic relationship set, and performing traffic anomaly detection and relationship dynamic update on the traffic scene semantic relationship set to obtain a time-varying traffic knowledge graph; performing multi-source information fusion and traffic driving knowledge reasoning on the time-varying traffic knowledge graph and the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation.

[0029] In practical applications, the road network topology features and vehicle trajectory features are first extracted from the multi-source traffic reorganization data, that is, the road network topology features are extracted by using the graph attention network (GAT), and the vehicle trajectory features are extracted by using the long short-term memory network (LSTM) to capture the spatial structure characteristics of the road network and the temporal characteristics of vehicle movement, and the initial road network and vehicle feature set are obtained. The continuous wavelet transform (CWT) is used to perform multi-scale analysis on the initial road network and vehicle feature set, and the line simplification algorithm is used for adaptive segmentation to identify the multi-scale characteristics of traffic flow, that is, to obtain multi-scale traffic flow characteristics, including intra-day changes, intra-week changes, and seasonal changes. Then, the hierarchical clustering algorithm is combined with the geographically weighted regression (GWR) to perform segment-level and regional-level spatial aggregation on the multi-scale traffic flow characteristics. The distance calculation of the hierarchical clustering adopts the dynamic time warping (DTW) distance, and its formula is: ; Among them, X and Y are two time series, is the path weight, d is the distance metric function, is a point in the sequence X With the points in sequence Y The distance between them is obtained through this processing, which not only reflects the characteristics of the local road section, but also reflects the overall traffic mode of the region. The multi-level traffic state representation is subjected to a long-range dependency analysis guided by the road network topology to capture the spatial dependency of the road network and establish long-range temporal dependency, so as to obtain the global traffic correlation characteristics, which not only considers the spatial structure of the road network, but also captures the changes in traffic modes over a long time scale. Then, based on the preset traffic scene events, the global traffic correlation characteristics are dynamically screened and fused, that is, the relevant features are dynamically screened according to different traffic scene events, and the gated recurrent unit is used to fuse the timing information to obtain the traffic scene representation. The knowledge graph is used to construct the traffic knowledge graph and perform rule reasoning on the traffic scene representation, and the probability soft logic is combined to deal with uncertainty and fuzziness, so as to transform the logical rules into a continuous value optimization problem, so as to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation.

[0030] Secondly, a multi-resolution graph matching algorithm and dynamic time warping (DTW) are used to perform multi-scale matching of the road network structure and the vehicle trajectory of the traffic feature vector set, that is, a quadtree structure is used to decompose the road network at multiple scales, and then a graph matching algorithm, such as the Hungarian algorithm or a graph matching network based on deep learning, is applied at each scale. At the same time, the DTW algorithm is used to find the optimal alignment path between the two time series through dynamic programming to align the vehicle trajectory in time, and the association mapping between the road network and the vehicle is obtained. The association mapping is then subjected to spatiotemporal consistency constraints and dynamic smoothing, that is, the vehicle trajectory is smoothed by a Kalman filter, and the conditional random field (CRF) is used to establish the spatial relationship between the road network structure and the vehicle trajectory, so as to obtain the spatiotemporal consistent traffic element characteristics; and then a hierarchical graph neural network is used to construct a hierarchical traffic scene graph for the traffic element characteristics and the preliminary dynamic traffic knowledge representation, so as to abstract and aggregate the graph structures corresponding to the traffic element characteristics and the preliminary dynamic traffic knowledge representation at different levels, so as to obtain a multi-level traffic semantic network, and based on the preset traffic behavior Based on the driving regulations, the graph attention network (GAT) and relational reasoning network (RRN) are used to perform contextual reasoning and correlation analysis of traffic rules on the multi-level traffic semantic network to obtain the traffic semantic association strength matrix; then based on the preset historical traffic pattern, the recurrent neural network (RNN) and conditional random field (CRF) are used to iteratively reason and detect the semantic description of the traffic semantic association strength matrix to capture the temporal dependency and establish the dependency between the label sequences, and a set of traffic scene semantic relations is generated. The variational autoencoder (VAE) and knowledge graph embedding technology are used to detect traffic anomalies and dynamically update the relations of the traffic scene semantic relations, and the results are fused with the preliminary dynamic traffic knowledge representation. Among them, VAE is used to learn the latent variable representation of the scene, and the knowledge graph embedding is used to encode the traffic knowledge into a low-dimensional vector, so as to obtain the time-varying traffic knowledge graph and the enhanced dynamic knowledge representation, which not only contains rich traffic feature information, but also integrates traffic rule knowledge and historical patterns, such as real-time updating of road network congestion status, and inferring the optimal traffic control strategy.

[0031] 104. Based on enhanced dynamic knowledge representation, deep probability fusion and uncertainty modeling are performed on multi-source traffic reorganization data to obtain traffic perception probability distribution, and a high-precision dynamic semantic map is constructed based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

[0032] In this embodiment, a multi-level traffic network probability graph is constructed for the enhanced dynamic knowledge representation to obtain a hierarchical traffic state probability distribution, and based on preset historical traffic data, a temporal and spatial consistency comparison is performed on the hierarchical traffic state probability distribution and the weight adjustment of the corresponding preliminary dynamic traffic knowledge representation at each level is performed to obtain the data confidence weight of the corresponding preliminary dynamic traffic knowledge representation at each level; the data confidence weight and the hierarchical traffic state probability distribution are Bayesian fused to obtain an updated traffic state posterior probability, and the traffic state posterior probability is subjected to temporal and spatial correlation analysis and Markov random field modeling to generate a dynamic traffic system evolution model; the dynamic traffic system evolution model is subjected to multi-objective traffic optimization solution to obtain a traffic state prediction and a traffic control strategy set, and the traffic state prediction and the traffic control strategy set are verified and updated with real-time traffic data to obtain a traffic perception probability distribution; the traffic perception probability distribution is subjected to a multi-level traffic network topology structure. The structure is extracted to obtain a preliminary traffic semantic skeleton diagram, and the preliminary traffic semantic skeleton diagram is subjected to spatiotemporal interpolation and dynamic feature completion to obtain a continuous multi-dimensional traffic scene representation; based on the traffic perception probability distribution, the continuous multi-dimensional traffic scene representation is subjected to probability evaluation and uncertainty area identification to obtain an area identification result, and based on the time-varying traffic knowledge graph, the uncertainty area is semantically supplemented to obtain a semantic supplement result; based on the semantic supplement result, the continuous multi-dimensional traffic scene representation is subjected to consistency constraint and map integration to generate a dynamic traffic semantic map, and the dynamic traffic semantic map is subjected to traffic flow change detection and local update to generate a time-varying traffic network model; the time-varying traffic network model is subjected to multi-scale traffic feature extraction and representation scheme optimization to obtain an adaptive multi-resolution traffic map, and the adaptive multi-resolution traffic map is subjected to spatiotemporal index construction and evolution analysis to obtain a final multi-resolution map representation of the fusion of the holographic road network and the Internet of Vehicles.

[0033] In practical applications, firstly, a hierarchical Bayesian network (HBN) is used to construct a multi-level traffic network probability graph for enhanced dynamic knowledge representation. That is, in HBN, each node represents a traffic state variable, and the edge represents the conditional dependency between variables. For each node Xi, its conditional probability distribution is P(Xi | Pa(Xi)), where Pa(Xi) is the parent node set of Xi. By defining a suitable conditional probability table to capture the dependencies at different levels in the traffic network, a hierarchical traffic state probability distribution is obtained. Based on the preset historical traffic data, the spatiotemporal kernel density estimation (ST-KDE) and adaptive weight learning algorithm are used to compare the spatiotemporal consistency and adjust the weight of the hierarchical traffic state probability distribution. That is, the spatiotemporal distribution of historical data is estimated by ST-KDE, and the Kullback-Leibler (KL) divergence is used to measure the consistency of the current state distribution with the historical distribution, so as to adjust the weights of the corresponding preliminary dynamic traffic knowledge representation at each level to obtain the data confidence weight. Then, the variational Bayesian inference method is used to perform Bayesian fusion on the data confidence weight and the hierarchical traffic state probability distribution. Through iterative optimization of Bayesian fusion, the updated traffic state posterior probability is obtained, and the spatiotemporal correlation analysis and Markov random field modeling are performed on the traffic state posterior probability. The spatiotemporal correlation analysis adopts Moran's I index is used to achieve this, and Markov random fields are used to generate dynamic traffic system evolution models to capture the spatiotemporal dependencies of traffic states, thereby generating dynamic traffic system evolution models; deep reinforcement learning combined with multi-objective optimization algorithms are then used to perform multi-objective traffic optimization on the dynamic traffic system evolution model, thereby obtaining traffic state predictions and traffic control strategy sets, and random gradient descent methods and extended Kalman filters are used to verify and update traffic state predictions and traffic control strategy sets with real-time traffic data, and finally traffic perception probability distribution is obtained. This distribution integrates multi-level and multi-dimensional traffic information and can accurately reflect the current and future traffic states. For example, it can simultaneously represent the current congestion level of a certain road section, the probability of a change in congestion status within the next 30 minutes, and the potential impact of this change on the surrounding road network.

[0034] Secondly, the graph clustering algorithm combined with spectral embedding technology is used to extract the multi-level traffic network topology structure of traffic perception probability distribution. That is, first construct the graph Laplacian matrix L=D - W, where D is the degree matrix and W is the adjacency matrix, then calculate the eigenvalues ​​and eigenvectors of L, select the eigenvectors corresponding to the first k smallest non-zero eigenvalues ​​to form a low-dimensional representation, and use the K-means algorithm to cluster the low-dimensional representation to obtain a preliminary traffic semantic skeleton graph, and then perform spatiotemporal interpolation and dynamic feature completion on the preliminary traffic semantic skeleton graph, that is, organize the traffic data in the preliminary traffic semantic skeleton graph into a third-order tensor X∈ R^(I×J×K), where I, J, and K represent the space, time, and feature dimensions, respectively, and then use Tucker decomposition to decompose X into a core tensor G and factor matrices A, B, and C. The decomposition formula is: X ≈ G ×1 A ×2 B ×3 C, and for missing values, use deep neural networks to predict and fill in, so as to obtain a continuous multi-dimensional traffic scene representation; then based on the traffic perception probability distribution, the continuous multi-dimensional traffic scene representation is probability evaluated and the uncertainty area is identified, that is, the variance of the predicted distribution is calculated to identify the area with high uncertainty, and the area identification result is obtained, and for the identified uncertainty area, semantic supplementation is performed based on the time-varying traffic knowledge graph to adaptively assign weights to different nodes to obtain the semantic supplementation result; then based on the semantic supplementation result, the continuous multi-dimensional traffic scene representation is constrained by consistency and map integration, that is, conditional random fields are used to achieve global consistency optimization, and the optimal label configuration is obtained by minimizing the energy function, and the optimized results are fused with the existing map data to generate a dynamic traffic semantic map, and then the long short-term memory network (LSTM) is used in combination with the convolutional neural network (CNN) to detect and locally update the dynamic traffic semantic map, that is, through the LSTM Capture long-term temporal dependencies, while CNN extracts local spatial features. When a changed area is detected, it will be locally updated to generate a time-varying traffic network model; then use multi-resolution analysis (MRA) technology combined with an adaptive grid refinement algorithm to extract multi-scale traffic features and optimize the representation scheme of the time-varying traffic network model, that is, decompose the signal into approximations and details of different scales through MRA, and use adaptive grid refinement to dynamically adjust the grid density according to the complexity of local features, so as to obtain an adaptive multi-resolution traffic map, and then perform spatiotemporal index construction and evolution analysis on the adaptive multi-resolution traffic map. The spatiotemporal index construction adopts a hybrid structure of quadtree and R-tree to efficiently support multi-scale spatiotemporal queries, and the evolution analysis uses temporal pattern mining algorithms such as sequence pattern mining to discover the long-term change trend of traffic patterns, so as to finally obtain a multi-resolution map representation of the fusion of holographic road network and Internet of Vehicles. This final map representation not only contains rich semantic information and dynamic features, but also supports multi-scale queries and analysis.For example, the traffic flow distribution of the entire city can be observed at a macro scale, and the traffic flow details of a single intersection can be viewed at a micro scale. At the same time, this map representation also has a time dimension, which can trace back historical status or predict future trends. By integrating the static structural information of the holographic road network and the dynamic behavior data of the Internet of Vehicles, this map representation provides a comprehensive, accurate, and real-time decision-making basis for the intelligent transportation system, which can support a variety of application scenarios such as traffic flow optimization, accident prevention, and path planning.

[0035] In an embodiment of the present invention, adaptive multimodal preprocessing and spatiotemporal alignment are performed on the acquired multi-source heterogeneous data, and then the spatiotemporal aligned data is hierarchically restructured, multi-scale spatiotemporal features are extracted, and a preliminary dynamic traffic knowledge representation is constructed. Then, multi-level semantic association is performed on the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation. Then, based on the enhanced dynamic knowledge representation, deep probability fusion and uncertainty modeling are performed on the multi-source traffic reorganized data to obtain a traffic perception probability distribution, and then a high-precision dynamic semantic map is constructed to form a fused multi-resolution map representation. The present application enhances the ability to understand and express complex traffic scenes through multi-level processing and semantic association, and integrates probability models and uncertainty analysis, thereby improving the system's adaptability and prediction capabilities for dynamic traffic environments, achieving effective fusion of holographic road network and Internet of Vehicles data, and improving the comprehensiveness and accuracy of traffic feature data in the fusion results.

[0036] The above describes the method for integrating the holographic road network and the vehicle network in the embodiment of the present invention. The following describes the device for integrating the holographic road network and the vehicle network in the embodiment of the present invention. Figure 2 In one embodiment of the present invention, a holographic road network and vehicle network integration device includes: The preprocessing module 201 is used to obtain multi-source heterogeneous data collected by the target sensor from the entire road network and the vehicle network, and perform adaptive multi-modal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; The structural reorganization module 202 is used to perform spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and to perform hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data; The semantic association module 203 is used to extract multi-scale spatiotemporal features from the multi-source traffic reorganization data, obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and perform multi-level semantic association on the multi-source traffic reorganization data based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation; The map representation module 204 is used to perform deep probability fusion and uncertainty modeling on the multi-source traffic reorganization data based on the enhanced dynamic knowledge representation to obtain traffic perception probability distribution, and to construct a high-precision dynamic semantic map based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

[0037] In an embodiment of the present invention, adaptive multimodal preprocessing and spatiotemporal alignment are performed on the acquired multi-source heterogeneous data, and then the spatiotemporal aligned data is hierarchically restructured, multi-scale spatiotemporal features are extracted, and a preliminary dynamic traffic knowledge representation is constructed. Then, multi-level semantic association is performed on the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation. Then, based on the enhanced dynamic knowledge representation, deep probability fusion and uncertainty modeling are performed on the multi-source traffic reorganized data to obtain a traffic perception probability distribution, and then a high-precision dynamic semantic map is constructed to form a fused multi-resolution map representation. The present application enhances the ability to understand and express complex traffic scenes through multi-level processing and semantic association, and integrates probability models and uncertainty analysis, thereby improving the system's adaptability and prediction capabilities for dynamic traffic environments, achieving effective fusion of holographic road network and Internet of Vehicles data, and improving the comprehensiveness and accuracy of traffic feature data in the fusion results.

[0038] above Figure 2 The holographic road network and vehicle network fusion device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The holographic road network and vehicle network fusion device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0039] Figure 3 3 is a schematic diagram of the structure of a holographic road network and vehicle network fusion device provided by an embodiment of the present invention. The holographic road network and vehicle network fusion device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage devices) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 may be temporary storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the holographic road network and vehicle network fusion device 300. Furthermore, the processor 310 may be configured to communicate with the storage medium 330 to execute a series of instruction operations in the storage medium 330 on the holographic road network and vehicle network fusion device 300.

[0040] The holographic road network and vehicle network fusion device 300 may also include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that Figure 3 The structure of the holographic road network and vehicle network fusion device shown does not constitute a limitation on the holographic road network and vehicle network fusion device, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0041] The present invention also provides a device for integrating a holographic road network and an Internet of Vehicles. The computer device includes a memory and a processor. The memory stores computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor executes each step of the method for integrating a holographic road network and an Internet of Vehicles in the above-mentioned embodiments.

[0042] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes each step of the method for integrating the holographic road network with the Internet of Vehicles.

[0043] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0044] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0045] The present application can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0046] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for integrating holographic road network and vehicle network, characterized in that: The method for integrating the holographic road network with the Internet of Vehicles includes: Acquire multi-source heterogeneous data collected by target sensors from the entire road network and the Internet of Vehicles, and perform adaptive multi-modal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; Performing spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and performing hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data; Extracting multi-scale spatiotemporal features from the multi-source traffic reorganization data to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation, performing multi-level semantic association on the multi-source traffic reorganization data to obtain an enhanced dynamic knowledge representation; Based on the enhanced dynamic knowledge representation, the multi-source traffic reorganization data is subjected to deep probability fusion and uncertainty modeling to obtain a traffic perception probability distribution, and a high-precision dynamic semantic map is constructed based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

2. The method for integrating holographic road network and Internet of Vehicles according to claim 1, characterized in that: The step of performing spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data includes: Performing time series analysis and frequency domain decomposition on the multi-source heterogeneous data to obtain multi-frequency band time components, and performing window calculation of each frequency band and merging and sorting of time series on the multi-frequency band time components to obtain multi-resolution multi-source heterogeneous time series data; Performing spatial registration of multi-target sensors and multi-source clock synchronization on the multi-source heterogeneous time series data to obtain spatiotemporal reference multi-source data, and performing dynamic coordinate system conversion on the spatiotemporal reference multi-source data to obtain global coordinate multi-source data of a global reference system; The data streams of multi-target sensors are aligned for the global coordinate multi-source data to obtain a synchronous multi-source data stream, and a multi-dimensional spatiotemporal index corresponding to the synchronous multi-source data stream is constructed to obtain spatiotemporally aligned multi-source heterogeneous data.

3. The method for integrating holographic road network and Internet of Vehicles according to claim 1 is characterized in that: The step of performing hierarchical structural reorganization on the time-space aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data includes: Performing multimodal anomaly detection on the spatiotemporally aligned multi-source heterogeneous data to obtain multi-source anomaly labeled data, and performing context-aware data repair on the multi-source anomaly labeled data to obtain preliminarily repaired multi-source heterogeneous data; Performing adaptive multi-scale filtering on the preliminarily repaired multi-source heterogeneous data to obtain noise-suppressed multi-source heterogeneous data, and performing multi-modal data traffic scene segmentation and fusion on the noise-suppressed multi-source heterogeneous data to obtain a semantically segmented multi-source traffic scene; The key traffic elements are extracted and the traffic attributes are labeled for the semantically segmented multi-source traffic scene to obtain a multi-source traffic structured representation, and the spatiotemporal dynamic characteristics of the multi-source traffic structured representation are modeled to obtain multi-source traffic reorganization data.

4. The method for integrating holographic road network and vehicle network according to claim 1, characterized in that: The step of extracting multi-scale spatiotemporal features from the multi-source traffic reorganization data to obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation includes: Extracting road network topology features and vehicle trajectory features from the multi-source traffic reorganization data to obtain an initial road network and vehicle feature set, and performing time series segmentation and periodicity analysis on the initial road network and vehicle feature set to obtain multi-scale traffic flow features; Performing segment-level and regional-level spatial aggregation on the multi-scale traffic flow characteristics to obtain a multi-level traffic state representation, and performing a road network topology-guided long-range dependency analysis on the multi-level traffic state representation to obtain a global traffic correlation feature; Based on preset traffic scene events, the global traffic-related features are dynamically screened and fused to obtain traffic scene representations, and traffic knowledge graphs and rule reasoning are performed on the traffic scene representations to obtain traffic feature vector sets and preliminary dynamic traffic knowledge representations.

5. The method for integrating holographic road network and vehicle network according to claim 4 is characterized in that: The method of performing multi-level semantic association on the multi-source traffic reorganization data based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation includes: Performing multi-scale matching of the road network structure and the vehicle trajectory on the traffic feature vector set to obtain a road network and vehicle association mapping, and performing spatiotemporal consistency constraints and dynamic smoothing processing on the road network and vehicle association mapping to obtain traffic element features with fine feature alignment; A hierarchical traffic scene graph is constructed for the preliminary dynamic traffic knowledge representation of the traffic element characteristics to obtain a multi-level traffic semantic network, and based on the preset traffic driving specifications, the multi-level traffic semantic network is subjected to contextual reasoning and correlation analysis of traffic rules to obtain a traffic semantic association intensity matrix; Based on the preset historical traffic pattern, the traffic semantic association strength matrix is ​​iteratively inferred and the semantic description is detected to generate a traffic scene semantic relationship set, and the traffic anomaly detection and relationship dynamic update are performed on the traffic scene semantic relationship set to obtain a time-varying traffic knowledge graph; The time-varying traffic knowledge graph and the preliminary dynamic traffic knowledge representation are subjected to multi-source information fusion and traffic driving knowledge inference to obtain an enhanced dynamic knowledge representation.

6. The method for integrating holographic road network and vehicle network according to claim 5 is characterized in that: The method of performing deep probability fusion and uncertainty modeling on the multi-source traffic reorganization data based on the enhanced dynamic knowledge representation to obtain traffic perception probability distribution includes: Constructing a multi-level traffic network probability graph for the enhanced dynamic knowledge representation to obtain a hierarchical traffic state probability distribution, and based on preset historical traffic data, performing a temporal and spatial consistency comparison on the hierarchical traffic state probability distribution and adjusting the weights of the corresponding preliminary dynamic traffic knowledge representations at each level to obtain the data confidence weights of the corresponding preliminary dynamic traffic knowledge representations at each level; Performing Bayesian fusion on the data confidence weight and the hierarchical traffic state probability distribution to obtain an updated traffic state posterior probability, and performing spatiotemporal correlation analysis and Markov random field modeling on the traffic state posterior probability to generate a dynamic traffic system evolution model; The dynamic traffic system evolution model is solved by multi-objective traffic optimization to obtain traffic state prediction and traffic control strategy set, and the traffic state prediction and traffic control strategy set are verified and updated by real-time traffic data to obtain traffic perception probability distribution.

7. The method for integrating holographic road network and vehicle network according to claim 6, characterized in that: The method of constructing a high-precision dynamic semantic map based on the traffic perception probability distribution to obtain a fused multi-resolution map representation includes: Extracting a multi-level traffic network topology structure from the traffic perception probability distribution to obtain a preliminary traffic semantic skeleton diagram, and performing spatiotemporal interpolation and dynamic feature completion on the preliminary traffic semantic skeleton diagram to obtain a continuous multi-dimensional traffic scene representation; Based on the traffic perception probability distribution, probability evaluation is performed on the continuous multi-dimensional traffic scene representation and the uncertainty region is identified to obtain a region identification result, and based on the time-varying traffic knowledge graph, semantic supplementation is performed on the uncertainty region to obtain a semantic supplementation result; Based on the semantic supplementation result, the continuous multi-dimensional traffic scene representation is subjected to consistency constraints and map integration to generate a dynamic traffic semantic map, and the dynamic traffic semantic map is subjected to traffic flow change detection and local update to generate a time-varying traffic network model; The time-varying traffic network model is subjected to multi-scale traffic feature extraction and representation scheme optimization to obtain an adaptive multi-resolution traffic map, and the adaptive multi-resolution traffic map is subjected to spatiotemporal index construction and evolution analysis to obtain a final multi-resolution map representation of the fusion of the holographic road network and the Internet of Vehicles.

8. A holographic road network and vehicle network integration device, characterized in that: The holographic road network and vehicle network fusion device comprises: A preprocessing module is used to obtain multi-source heterogeneous data collected by the target sensor from the entire road network and the Internet of Vehicles, and to perform adaptive multi-modal preprocessing on the multi-source heterogeneous data to obtain preprocessed multi-source heterogeneous data; A structural reorganization module is used to perform spatiotemporal alignment on the preprocessed multi-source heterogeneous data to obtain spatiotemporal aligned multi-source heterogeneous data, and to perform hierarchical structural reorganization on the spatiotemporal aligned multi-source heterogeneous data to obtain multi-source traffic reorganized data; A semantic association module, used to extract multi-scale spatiotemporal features from the multi-source traffic reorganization data, obtain a traffic feature vector set and a preliminary dynamic traffic knowledge representation, and perform multi-level semantic association on the multi-source traffic reorganization data based on the traffic feature vector set and the preliminary dynamic traffic knowledge representation to obtain an enhanced dynamic knowledge representation; A map representation module is used to perform deep probability fusion and uncertainty modeling on the multi-source traffic reorganization data based on the enhanced dynamic knowledge representation to obtain a traffic perception probability distribution, and to construct a high-precision dynamic semantic map based on the traffic perception probability distribution to obtain a fused multi-resolution map representation.

9. A holographic road network and vehicle network integration device, characterized in that: The holographic road network and vehicle network integration device comprises: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory so that the holographic road network and vehicle network fusion device executes each step of the holographic road network and vehicle network fusion method as described in any one of claims 1-7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the various steps of the method for integrating the holographic road network and the Internet of Vehicles as described in any one of claims 1 to 7 are implemented.

Images