Data partition storage method, storage device, storage medium and product

Abstract

The invention provides a data partition storage method, a storage device, a storage medium and a product, and the method comprises the steps: obtaining traffic node data which at least comprises traffic node information and a connection relation between traffic nodes; constructing a structural similarity matrix according to the connection relationship between the traffic nodes, constructing a flow similarity matrix according to traffic flow data in the traffic node information, and constructing a joint similarity matrix according to the structural similarity matrix and the flow similarity matrix; and dividing the traffic nodes into a plurality of sub-clusters according to the joint similarity matrix. The technical problems that in the prior art, data centralized management is prone to being threatened by single-point faults and hostile attacks, data abuse and privacy disclosure are caused by lack of effective isolation of data, and data cannot be traced and verified are solved. By adopting the method provided by the invention, the data can be safely stored and managed, and the accuracy of traffic data partitioning is improved.

Description

Technical Field

[0001] The present invention relates to the field of vehicles, and in particular to a data partition storage method, storage device, storage medium and product. Background Art

[0002] In the existing technology, with the development of intelligent transportation systems, the vehicle-road-cloud integrated system has played an important role in improving traffic efficiency and enhancing road safety. The vehicle-road-cloud integrated system achieves real-time monitoring, information sharing and intelligent decision-making through the exchange of large amounts of data between vehicles, road infrastructure and cloud platforms.

[0003] However, the existing vehicle-road-cloud integrated system has the following problems in data management:

[0004] Security risks of centralized data management: Traditional vehicle-road-cloud integrated systems usually rely on centralized cloud platforms for data storage and management, which are vulnerable to single point failures and malicious attacks, leading to data leakage and system paralysis.

[0005] Lack of effective isolation in data sharing: Different types of data (such as vehicle location data, user personal information, road status information, etc.) are stored and processed on the same platform, lacking effective access control and isolation mechanisms, which can easily lead to data abuse and privacy leakage.

[0006] Data cannot be traced or verified: In traditional systems, the source and modification process of data are difficult to trace, and there is a lack of reliable verification methods, which affects the reliability of data and the accuracy of system decisions.

[0007] In the big data processing in the transportation field, the data volume is huge and dynamically changing, and the traditional static partitioning method is difficult to adapt to the real-time and diversified data requirements.

[0008] The above problems limit the further development of vehicle-road-cloud integrated systems in terms of security, privacy protection and data management efficiency. Summary of the invention

[0009] Based on the above problems, the data partition storage method, storage device, storage medium and product proposed in the present invention solve the technical problem of the centralized management of data in the prior art that is vulnerable to single point failure and malicious attack, solve the technical problem of data abuse and privacy leakage caused by the lack of effective isolation of data in the prior art, and solve the technical problem of data being untraceable and unverifiable in the prior art. The data partition storage method proposed in the present invention can enable data to be securely stored and managed, eliminate single point failures and improve the accuracy of traffic data partitioning.

[0010] The present invention provides a method for storing data in partitions, comprising:

[0011] A data acquisition step, for acquiring traffic node data of a plurality of traffic nodes, wherein the traffic node data at least includes traffic node information of the traffic nodes and connection relationships between the traffic nodes, and the traffic node information at least includes traffic flow data;

[0012] A similarity matrix construction step is used to construct a structural similarity matrix according to the connection relationship between the traffic nodes, to construct a flow similarity matrix according to the traffic flow data in the traffic node information, and to construct a joint similarity matrix according to the structural similarity matrix and the flow similarity matrix;

[0013] The spectral clustering division step is used to divide the plurality of traffic nodes into a plurality of sub-clusters according to the joint similarity matrix, and store the traffic node data corresponding to different sub-clusters into different storage areas.

[0014] In addition, constructing a structural similarity matrix according to the connection relationship between the traffic nodes includes:

[0015] An adjacency matrix between traffic nodes is constructed, and the adjacency matrix is ​​used as a structural similarity matrix. Elements in the adjacency matrix represent whether two traffic nodes are connected.

[0016] In addition, constructing a traffic similarity matrix according to the traffic flow data in the traffic node information includes:

[0017] The traffic similarity matrix is ​​constructed as:

[0018]

[0019] Among them, S f is the traffic similarity matrix, DTW is the similarity function, is the standard value of the traffic flow data of the ith traffic node, is the standard value of the traffic flow data of the jth traffic node.

[0020] In addition, constructing a joint similarity matrix according to the structural similarity matrix and the traffic similarity matrix includes:

[0021] The joint similarity matrix is ​​constructed as:

[0022] S=αS s +βS f ,

[0023] Among them, S is the joint similarity matrix, α is the first weight coefficient, β is the second weight coefficient, and α+β=1, S s is the structural similarity matrix, S f is the traffic similarity matrix.

[0024] In addition, dividing the plurality of traffic nodes into a plurality of sub-clusters according to the joint similarity matrix includes:

[0025] Calculating a joint Laplacian matrix according to the joint similarity matrix;

[0026] Solving the eigenvectors corresponding to the first K smallest eigenvalues ​​of the Laplace matrix, and forming a characteristic matrix with the eigenvectors corresponding to the first K smallest eigenvalues;

[0027] Each row of the feature matrix corresponds to a feature vector of the traffic node, and each column of the feature matrix represents a feature of the traffic node. Clustering calculation is performed on the feature matrix to cluster the multiple traffic nodes into different sub-clusters.

[0028] In addition, after the spectral clustering division step, a dynamic adjustment step is also included, and the dynamic adjustment step includes: executing the data acquisition step, the similarity matrix construction step and the spectral clustering division step once every preset time period.

[0029] In addition, the method further comprises an access verification step: when the storage area is accessed, the access rights and / or access rules of the visitor are verified, and the access rights and / or access rules of different storage areas are different.

[0030] The present invention also provides an electronic device, comprising:

[0031] at least one processor; and,

[0032] a memory communicatively connected to at least one of the processors; wherein,

[0033] The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors to enable the at least one processor to execute the data partition storage method as described in any one of the above items.

[0034] The present invention further proposes a storage medium, wherein the storage medium stores computer instructions, and when a computer executes the computer instructions, it is used to execute the data partition storage method as described in any one of the above items.

[0035] The present invention further proposes a computer program product, comprising a computer program / instruction, wherein the computer program / instruction, when executed by a processor, implements the data partition storage method as described in any one of the above items.

[0036] The data partition storage method proposed in the present invention solves the technical problem of the centralized management of data in the prior art that is vulnerable to single point failure and malicious attack, solves the technical problem of data abuse and privacy leakage caused by the lack of effective data isolation in the prior art, and solves the technical problem of data being untraceable and unverifiable in the prior art. The data partition storage method proposed in the present invention can enable data to be securely stored and managed, eliminate single point failures, and improve the accuracy of traffic data partitioning. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 A flowchart of a method for storing data partitions provided by an embodiment of the present invention;

[0038] Figure 2 A flowchart of a method for storing data partitions provided by an embodiment of the present invention;

[0039] Figure 3 A flowchart of a method for determining sub-clusters by re-dividing the clusters according to an embodiment of the present invention;

[0040] Figure 4 A schematic diagram of user access data provided by one embodiment of the present invention;

[0041] Figure 5 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0042] The present invention is further described in detail below in conjunction with specific embodiments and drawings. It is only intended to elaborate on the specific embodiments of the present invention and does not impose any limitation on the present invention. The protection scope of the present invention shall be subject to the claims.

[0043] The present invention is further described in detail below in conjunction with specific embodiments and drawings. It is only intended to elaborate on the specific embodiments of the present invention and does not impose any limitation on the present invention. The protection scope of the present invention shall be subject to the claims.

[0044] The professional terms involved in this invention are described as follows:

[0045] Smart Contract: An automatically executed contract code deployed on the blockchain to define and execute preset business logic and rules.

[0046] Consensus Mechanism: A protocol used in a blockchain network to reach consensus among all nodes in the network. Common ones include Proof of Work (PoW) and Proof of Stake (PoS).

[0047] Cryptographic Algorithm: A mathematical algorithm used to protect data security, including symmetric encryption, asymmetric encryption, and hash algorithms.

[0048] In the TL-DCRNN (Temporal Latent-Dynamic Graph Convolutional Recurrent Neural Network) model, the graph partitioning method for the highway network is mainly to improve the efficiency and prediction accuracy of the model. Dividing the entire highway network into different sub-clusters can help better model local traffic patterns and relationships.

[0049] Reference Figure 1 The present invention proposes a method for data partition storage, comprising:

[0050] The data acquisition step S001 is used to acquire traffic node data of a plurality of traffic nodes, wherein the traffic node data at least includes traffic node information of the traffic nodes and connection relationships between the traffic nodes, and the traffic node information at least includes traffic flow data;

[0051] The step S002 of constructing a similarity matrix is ​​used to construct a structural similarity matrix according to the connection relationship between the traffic nodes, to construct a flow similarity matrix according to the traffic flow data in the traffic node information, and to construct a joint similarity matrix according to the structural similarity matrix and the flow similarity matrix;

[0052] The spectral clustering division step S003 is used to divide the multiple traffic nodes into multiple sub-clusters according to the joint similarity matrix, and store the traffic node data corresponding to different sub-clusters in different storage areas.

[0053] In the data acquisition step S001, optionally, the following processing is performed:

[0054] Define the traffic network graph G = (V, E), where: C represents a traffic node (such as an intersection, a road section),

[0055] E represents the connection relationship between traffic nodes.

[0056] Collect time-series traffic flow data:

[0057] For each traffic node v i ∈V, collect its time series traffic flow data X i =[x i1 ,x i2 ,…,x iT ], where x it Represents a traffic node v i Traffic flow at time t.

[0058] Traffic flow data is standardized to eliminate the impact of different magnitudes:

[0059]

[0060] Among them, μ i and σ i They are traffic nodes v i The mean and standard deviation of the flow rate.

[0061] In the step S002 of constructing a similarity matrix, when constructing a structural similarity matrix according to the connection relationship between traffic nodes, the following method can be used:

[0062] When constructing the structural similarity matrix, the adjacency matrix between the traffic nodes is first constructed and the adjacency matrix is ​​used as the structural similarity matrix. The elements in the adjacency matrix indicate whether two traffic nodes are connected.

[0063] Define the adjacency matrix A:

[0064] Suppose we have an undirected graph consisting of 5 nodes. The adjacency matrix A of this graph is as follows:

[0065]

[0066] Among them, A[i][j]=1 indicates that there is an edge between traffic node i and traffic node j, and A[i][j]=0 indicates that there is no edge between traffic node i and traffic node j.

[0067] In this example, the adjacency matrix is ​​represented as follows:

[0068] Traffic node 0 is connected to traffic nodes 1 and 2.

[0069] Traffic node 1 is connected to traffic nodes 0, 2, and 3.

[0070] Traffic node 2 is connected to traffic nodes 0, 1, and 4.

[0071] Traffic node 3 is connected to traffic nodes 1 and 4.

[0072] Traffic node 4 is connected to traffic nodes 2 and 3.

[0073] By constructing an adjacency matrix, we can obtain a matrix showing whether the traffic nodes are connected, in preparation for subsequent calculations.

[0074] Then construct the traffic similarity matrix S f

[0075] The similarity function is used to calculate the flow similarity between traffic nodes. Preferably, the dynamic time warping (DTW) algorithm is used to calculate the flow similarity between traffic nodes, and the flow similarity matrix S is defined. f :

[0076]

[0077] The smaller the distance calculated by DTW, the closer the traffic node v is. i and v j The more similar the traffic flow patterns are.

[0078] Dynamic Time Warping (DTW) is an algorithm for measuring the similarity between two time series data sequences, especially for sequences that may have nonlinear deformations on the time axis. DTW overcomes the technical problem of mismatched time axes by finding the best alignment between two sequences to calculate the distance between them. In traffic data analysis, DTW can be used to compare traffic data from different nodes, even if the time distribution of these data is inconsistent.

[0079] The traffic similarity matrix is ​​constructed by calculating the similarity of traffic time series data between each pair of traffic nodes. DTW is usually used to construct the matrix. The elements of the matrix represent the traffic similarity between the corresponding traffic nodes. Generally, the smaller the distance value, the more similar the traffic patterns between the traffic nodes.

[0080] Example:

[0081] Assume that there are three traffic nodes (A, B, C), and their traffic time series data are as follows:

[0082] Traffic node A: 5, 10, 15, 20

[0083] Traffic node B: 5, 12, 15, 18

[0084] Traffic node C: 3, 9, 14, 22

[0085] We use DTW to calculate the traffic similarity between traffic nodes. For example:

[0086] DTW(A,B), by aligning the sequences 5,10,15,20 and 5,12,15,18, the DTW distance is calculated as d AB .

[0087] DTW(A,C), by aligning sequences 5,10,15,20 and 3,9,14,22, the DTW distance is calculated as d AC .

[0088] DTW(B,C), by aligning sequences 5,12,15,18 and 3,9,14,22, the DTW distance is calculated as d BC .

[0089] The DTW algorithm can be adjusted according to the specific situation. Here, it is assumed that the calculated distance value is:

[0090] d AB =3

[0091] d AC =6

[0092] d BC =5

[0093] Based on these calculation results, the traffic similarity matrix S f It can be expressed as:

[0094]

[0095] S f [i][j] represents the traffic similarity between traffic node i and traffic node j (measured by DTW distance), and the diagonal elements are 0 (the distance between the traffic node and itself).

[0096] The flow relationship between traffic nodes is:

[0097] If S f [0][1]=3 (flow similarity between traffic node A and traffic node B). A smaller distance indicates that the flow patterns of these two traffic nodes are relatively similar.

[0098] If S f [0][2]=6 (flow similarity between traffic node A and traffic node C), a larger distance indicates that the flow patterns of these two traffic nodes are relatively less similar.

[0099] Such a traffic similarity matrix can not only help identify traffic nodes with similar traffic patterns, but also provide a basis for subsequent graph partitioning and traffic flow prediction. DTW, as the core similarity calculation method, can effectively capture the nonlinear characteristics and dynamic changes of traffic data between nodes.

[0100] Optionally, traffic flow prediction is performed based on the traffic flow similarity between traffic nodes provided by the traffic flow similarity matrix. Optionally, if the similarity between the traffic flow time series data of traffic node A and the traffic flow time series data of traffic node B is less than a preset threshold, the traffic flow time series data of traffic node B can be predicted based on the traffic flow time series data of traffic node A in a subsequent period of time.

[0101] When constructing a joint similarity matrix based on the structural similarity matrix and the traffic similarity matrix, the following methods can be used:

[0102] S=αS s +βS f ,

[0103] Among them, α and β are weight coefficients, satisfying α+β=1, S s is the structural similarity matrix, S f is the traffic similarity matrix. α and β are used to balance the importance of the structure similarity matrix and the traffic similarity matrix.

[0104] In the spectral clustering step S003, the traffic nodes are divided into multiple sub-clusters according to the joint similarity matrix, and the traffic node data corresponding to different sub-clusters are stored in different storage areas. The optional methods are:

[0105] Calculate the joint Laplacian matrix based on the joint similarity matrix;

[0106] Solve the eigenvectors corresponding to the first K smallest eigenvalues ​​of the Laplace matrix. The eigenvectors corresponding to the first K smallest eigenvalues ​​form a characteristic matrix.

[0107] Each row of the feature matrix corresponds to a feature vector of a traffic node, and each column of the feature matrix represents a feature of a traffic node. Clustering calculation is performed on the feature matrix to cluster multiple traffic nodes into different sub-clusters.

[0108] Optionally, a K-means algorithm is selected to perform clustering calculation on the feature matrix.

[0109] The feature matrix is ​​denoted as U. The feature matrix is ​​obtained by extracting features from traffic nodes, where each row represents a feature vector of a traffic node. The feature vector contains information on multiple dimensions related to traffic flow and node structure, usually including: flow mean, flow standard deviation, number of road connections, flow pattern characteristics, etc.

[0110] The structure of the feature matrix is ​​as follows:

[0111] The feature matrix U is set to 5×n, where n represents the dimension of the feature. Assuming n=4, the feature matrix can be expressed as:

[0112]

[0113] Here, it is assumed that the meaning of each feature is as follows:

[0114] First dimension: traffic mean (number of vehicles / hour)

[0115] Second dimension: traffic standard deviation (number of vehicles / hour)

[0116] The third dimension: the number of road connections (the number of roads directly connected to this node)

[0117] Fourth dimension: traffic pattern characteristics (scores based on historical patterns)

[0118] The K-means clustering steps are as follows:

[0119] Select K value: Set the number of clusters K, assuming K = 2.

[0120] Initialization: Randomly select K nodes as the initial centroids, for example, select the first and second rows:

[0121] C1=[0.5,20,3,0.1],C2=[0.7,18,4,0.2]

[0122] Calculate distance: Calculate the distance from each node to the centroid (Euclidean distance can be used here).

[0123] Assign clusters: Based on distance, assign each node to the centroid that is closest to it.

[0124] Update centroids: Calculate new centroids for each cluster's nodes.

[0125] Iteration: Repeat the process of calculating distances and updating clusters until the centroids no longer change.

[0126] The specific examples are as follows:

[0127] Assume that after K-means clustering, the following sub-cluster assignments are obtained:

[0128] Subcluster 1: Node 1 (0.5, 20, 3, 0.1), Node 3 (0.4, 25, 2, 0.05), Node 4 (0.6, 15, 3, 0.15)

[0129] Subcluster 2: Node 2 (0.7, 18, 4, 0.2), Node 5 (0.8, 22, 5, 0.1)

[0130] Subcluster structure analysis:

[0131] Subcluster 1: All traffic nodes in this subcluster have similar mean traffic flow values ​​(e.g., 0.4-0.6) and fewer connections, representing road nodes with low traffic flow.

[0132] Subcluster 2: Node characteristics show higher traffic mean (e.g., 0.7-0.8) and number of connections, representing high-traffic traffic nodes.

[0133] The clustering obtained in this way can assist traffic management systems to better understand and respond to road conditions with different traffic patterns.

[0134] The present invention proposes a data partition isolation method, which aims to solve the security and data management problems in the prior art through decentralized data management and an effective partition isolation mechanism. The distributed storage and management of data in the present invention eliminates single points of failure and improves the robustness and security of the system; the present invention can ensure the authenticity and integrity of the data. The improved traffic data partitioning algorithm based on TL-DCRNN combines the structural characteristics of the traffic network with the dynamic characteristics of the traffic flow, and dynamically divides the traffic network into multiple highly cohesive and low-coupled sub-cluster partitions, thereby improving the accuracy of traffic data partitioning.

[0135] The data partition storage method proposed in the present invention solves the technical problem of the centralized data management in the prior art that is vulnerable to single point failure and malicious attack, solves the technical problem of data abuse and privacy leakage caused by the lack of effective data isolation in the prior art, and solves the technical problem of data untraceability and unverification in the prior art. The data partition storage method proposed in the present invention can make data securely stored and managed, eliminate single point failures and improve the accuracy of traffic data partitioning; and can allocate subsequent traffic flow according to the traffic node data stored in the partition, so that the traffic network is smoother and congestion is reduced; it can accurately obtain the partitioned traffic node data corresponding to the traffic accident through information such as geographic location, so that the accountability for traffic accidents is easier, more accurate and faster.

[0136] In one embodiment, constructing a structural similarity matrix according to the connection relationship between traffic nodes includes:

[0137] An adjacency matrix between traffic nodes is constructed and used as a structural similarity matrix. The elements in the adjacency matrix indicate whether two traffic nodes are connected.

[0138] Define the adjacency matrix A:

[0139] Suppose we have an undirected graph consisting of 5 nodes. The adjacency matrix A of this graph is as follows:

[0140]

[0141] Among them, A[i][j]=1 indicates that there is an edge between traffic node i and traffic node j, and A[i][j]=0 indicates that there is no edge between traffic node i and traffic node j.

[0142] In this example, the adjacency matrix is ​​represented as follows:

[0143] Traffic node 0 is connected to traffic nodes 1 and 2.

[0144] Traffic node 1 is connected to traffic nodes 0, 2, and 3.

[0145] Traffic node 2 is connected to traffic nodes 0, 1, and 4.

[0146] Traffic node 3 is connected to traffic nodes 1 and 4.

[0147] Traffic node 4 is connected to traffic nodes 2 and 3.

[0148] By constructing an adjacency matrix, we can obtain a matrix showing whether the traffic nodes are connected, in preparation for subsequent calculations.

[0149] In one embodiment, constructing a traffic similarity matrix according to traffic flow data in traffic node information includes:

[0150] The traffic similarity matrix is ​​constructed as:

[0151]

[0152] Among them, S f is the traffic similarity matrix, DTW is the similarity function, is the standard value of the traffic flow data of the ith traffic node, is the standard value of the traffic flow data of the jth traffic node.

[0153] The DTW similarity function is preferably a dynamic time planning algorithm function.

[0154] DTW, as the core similarity calculation method, can effectively capture the nonlinear characteristics and dynamic changes of traffic data between nodes.

[0155] Optionally, traffic flow prediction is performed based on the traffic flow similarity between traffic nodes provided by the traffic flow similarity matrix. Optionally, if the similarity between the traffic flow time series data of traffic node A and the traffic flow time series data of traffic node B is less than a preset threshold, the traffic flow time series data of traffic node B can be predicted based on the traffic flow time series data of traffic node A in a subsequent period of time.

[0156] In one embodiment, constructing a joint similarity matrix according to the structure similarity matrix and the flow similarity matrix includes:

[0157] The joint similarity matrix is ​​constructed as:

[0158] S=αS s +βS f ,

[0159] Among them, S is the joint similarity matrix, α is the first weight coefficient, β is the second weight coefficient, and α+β=1, S s is the structural similarity matrix, S f is the traffic similarity matrix. α and β are used to balance the importance of the structure similarity matrix and the traffic similarity matrix.

[0160] By constructing a joint similarity matrix to balance the proportions of the structural similarity matrix and the flow similarity matrix, the weight proportions of the two meet the requirement of summing to 1, thereby obtaining a relatively objective similarity between traffic nodes.

[0161] In one embodiment, dividing the traffic node into a plurality of sub-clusters according to the joint similarity matrix includes:

[0162] Calculate the joint Laplacian matrix based on the joint similarity matrix;

[0163] Solve the eigenvectors corresponding to the first K smallest eigenvalues ​​of the Laplace matrix. The eigenvectors corresponding to the first K smallest eigenvalues ​​form a characteristic matrix.

[0164] The mean algorithm is applied to each row of the feature matrix, and the mean of each row corresponds to a subcluster.

[0165] The joint Laplacian matrix is ​​calculated as follows:

[0166] According to the joint similarity matrix S, calculate the joint Laplacian matrix L:

[0167] L = DS,

[0168] Where D is the degree matrix, defined as:

[0169]

[0170] The eigenvector is calculated as follows:

[0171] Solve the eigenvectors corresponding to the first K smallest eigenvalues ​​of the Laplace matrix L to form the characteristic matrix U∈R |V|×K .

[0172] The K-means clustering calculation is as follows:

[0173] Apply the K-means algorithm to each row of the feature matrix U and divide the traffic nodes into K subclusters {C1, C2, …, C K}.

[0174] By dividing the traffic nodes into multiple sub-clusters according to the joint similarity matrix, different traffic nodes are classified into different sub-clusters and stored in different storage areas, that is, data is stored in a data partition isolation manner. Data partition isolation refers to a method in which data is divided and isolated according to different categories, permissions or usage scenarios in an information system to improve data security and management efficiency. Data is stored in different categories, such as vehicle data, road infrastructure upload data, and cloud data.

[0175] Reference Figure 2 In one embodiment, the spectral clustering division step further includes a dynamic adjustment step, and the dynamic adjustment step includes: executing the data acquisition step, the similarity matrix construction step and the spectral clustering division step once every preset time period.

[0176] The dynamic adjustment step is used to dynamically adjust the division of sub-clusters every preset time period.

[0177] When monitoring traffic flow, time window division is required to divide the time series traffic data into multiple time windows (for example, the time window is hourly or daily), that is, the traffic flow is obtained once every preset time period to capture the dynamic changes of traffic flow.

[0178] Update the traffic similarity matrix:

[0179] Recalculate the traffic similarity matrix L periodically (such as every hour or every day) f :

[0180]

[0181] Reconstruct the joint similarity matrix:

[0182] According to the new traffic similarity, update the joint similarity matrix S:

[0183]

[0184] Re-perform spectral clustering:

[0185] Based on the updated joint similarity matrix S new , recalculate the Laplace matrix L new =D new -S new .

[0186] Solve for the new characteristic matrix U new , and apply K-means algorithm to adjust the sub-cluster division

[0187]

[0188] Optionally, if a significant traffic pattern change is detected (eg, by monitoring the rate of change of the clustering results), a re-division of sub-clusters is triggered, ie, a dynamic adjustment step is performed to adapt to the latest traffic pattern.

[0189] Suppose we analyze road traffic flow data in a city traffic management system and monitor the traffic flow patterns in two time periods. Our goal is to divide the traffic flow data of different road sections into multiple subclusters through cluster analysis, and judge whether the change of traffic flow pattern is significant based on the change of cluster center.

[0190] Optionally, refer to Figure 3 ,The steps to judge the change of traffic flow pattern according to the change of cluster center are as follows:

[0191] First, two different traffic nodes are obtained, and the traffic flow data of the two traffic nodes at the first time point and the second time point are obtained respectively;

[0192] Secondly, the traffic flow data of the two traffic nodes are clustered at the first time point and the second time point to obtain the first time point sub-cluster and the second time point sub-cluster corresponding to the two traffic nodes respectively;

[0193] Next, the distance between the first time point sub-cluster and the second time point sub-cluster of each traffic node is calculated, for example, the Euclidean algorithm is used to calculate the Euclidean distance between the two, and the sum of the distances of the two traffic nodes is calculated according to the distance of each traffic node, which is recorded as the total change distance of the cluster center. Other methods for calculating the distance may include Manhattan distance, Chebyshev distance, etc.

[0194] Finally, the initial mean of the cluster center is calculated based on the traffic flow data of the traffic nodes, and the value of (total change distance of the cluster center / initial mean of the cluster center)*100% is used as the traffic flow change rate; if the traffic flow change rate exceeds the preset threshold, such as 10%, it is considered that the traffic flow change is significant and the sub-clusters need to be re-divided.

[0195] The following example illustrates this monitoring and calculation process:

[0196] Data preparation:

[0197] At time points t1 and t2, we obtain periodic traffic flow data on different road sections. The traffic flow of each road section is recorded in units of vehicles per hour and divided into 2 clusters by K-means clustering.

[0198] At time t1:

[0199] The traffic flow data of the road section is clustered, and the following cluster centers are obtained:

[0200] Cluster center C(t1):

[0201] Cluster 1: (traffic volume is 30 vehicles / hour, speed is 100km / h),

[0202] Cluster 2: (Traffic volume is 80 vehicles / hour and speed is 60km / h).

[0203] At time t2:

[0204] The traffic flow data collected at the next time point is used to obtain the new cluster center:

[0205] Cluster center C(t2):

[0206] Cluster 1: (Traffic volume is 35 vehicles / hour, speed is 95km / h)

[0207] Cluster 2: (Traffic volume is 90 vehicles / hour, speed is 50km / h)

[0208] Calculate the rate of change:

[0209] Calculate the change between two cluster centers and calculate the Euclidean distance for each cluster at different time points:

[0210]

[0211] Calculate the total change distance of cluster centers:

[0212] Total change distance:

[0213] Calculate the initial means of the cluster centers:

[0214]

[0215] Calculate the rate of change:

[0216]

[0217] Results and Analysis:

[0218] According to calculations, the change rate of the cluster center from time point t1 to t2 is about 20.55%.

[0219] According to a preset threshold (e.g., 10%), this change can be considered significant, indicating that the traffic flow pattern has changed significantly, so it may be necessary to adjust the traffic management strategy or re-divide the subclusters.

[0220] Optionally, the process of triggering the re-sub-clustering starts with calculating the traffic similarity, followed by updating the joint similarity matrix and performing new clustering in the spectral clustering step. This process ensures that the system can adapt to the latest traffic patterns.

[0221] Sub-clustering results:

[0222] Finally, we get K subclusters, each of which is C kThe nodes in the cluster have high structural similarity and traffic similarity, and the data sub-clusters are partitioned and stored. Each sub-cluster is a sub-graph of the TL-DCRNN model, and can independently model the spatiotemporal characteristics and predict the traffic flow in the future, improving the efficiency and accuracy of the overall prediction.

[0223] In one embodiment, each sub-cluster is taken as a sub-graph, and spatiotemporal feature modeling and traffic node flow prediction are performed based on the sub-graph.

[0224] In one embodiment, the method further comprises an access verification step: when a storage area is accessed, the access rights and / or access rules of the visitor are verified, and the access rights and / or access rules of different storage areas are different.

[0225] Optionally, the storage area is stored using blockchain technology.

[0226] Blockchain: A decentralized distributed ledger technology that uses encryption algorithms, consensus mechanisms and chain data structures to achieve data immutability, traceability and trustless management.

[0227] Optionally, before the traffic node data is partitioned and stored, the data is encrypted and hashed to ensure the security and integrity of the data.

[0228] Reference Figure 4 , optionally, control the data on-chain and access rights through smart contracts. By defining smart contracts, the access rules and permission management mechanisms for different data storage areas are specified. When the storage area is accessed, the access rights and / or access rules of the visitor are verified. When the user requests to access the data, the smart contract automatically verifies the permission and grants or denies access.

[0229] Optionally, adopt an improved proof-of-stake consensus mechanism to improve the efficiency and security of the blockchain network. Introduce a data verification step to ensure the legitimacy and accuracy of the data on the chain.

[0230] Proof of Stake (PoS) is a blockchain consensus mechanism, the core of which is to determine the user's eligibility to participate in block verification and reward distribution by the amount of coins held and the time they hold coins. Unlike traditional proof of work (PoW), PoS does not rely on a large amount of computing resources, but uses "held equity" as weight. This mechanism reduces energy consumption while encouraging coin holders to actively participate in the security and operation of the network.

[0231] Use smart contracts to define and execute business logic to achieve automated data management and access control. Use smart contracts to automate data sharing and transactions, and improve system operation efficiency.

[0232] The design of efficient data verification and consensus mechanism is suitable for the efficient consensus mechanism of vehicle-road-cloud integration scenario, which improves the performance and scalability of blockchain network. The data verification mechanism is introduced to ensure the legitimacy and correctness of data before entering the blockchain.

[0233] Reference Figure 5 The present invention also provides a hardware structure diagram of an electronic device, including:

[0234] Figure 5 A processor 501 is taken as an example.

[0235] The electronic device is preferably a controller of the vehicle. The electronic device may further include: an input device 505 and a display device 504 .

[0236] The processor 501, the memory 502, the input device 505 and the display device 504 may be connected via a bus or other means, and the figure takes the connection via a bus as an example.

[0237] The memory 502 is a non-volatile computer-readable storage medium that can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules corresponding to the data partition storage method in the embodiment of the present application, for example, Figure 2 The processor 501 executes various functional applications and data processing by running the non-volatile software programs, instructions and modules stored in the memory 502, that is, the method for data partition storage in the above embodiment is implemented.

[0238] The memory 502 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required by at least one function; the data storage area may store data created according to the use of the data partition storage method, etc. In addition, the memory 502 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 502 may optionally include a memory remotely arranged relative to the processor 501, and these remote memories may be connected to a device for executing the data partition storage method via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0239] The input device 505 can receive user clicks and generate signal inputs related to user settings of the data partition storage method and function control. The display device 504 can include display devices such as a display screen.

[0240] One or more modules are stored in the memory 502, and when executed by one or more processors 501, the method for data partition storage in any of the above method embodiments is executed.

[0241] The data partition storage method proposed in the present invention solves the technical problem of the centralized data management in the prior art that is vulnerable to single point failure and malicious attack, solves the technical problem of data abuse and privacy leakage caused by the lack of effective data isolation in the prior art, and solves the technical problem of data being untraceable and unverifiable in the prior art. The data partition storage method proposed in the present invention can enable data to be securely stored and managed, eliminate single point failures, and improve the accuracy of traffic data partitioning.

[0242] An embodiment of the present invention provides a storage medium, which stores computer instructions. When a computer executes the computer instructions, it is used to execute all steps of the data partition storage method as described above.

[0243] In the context of the present disclosure, a storage medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or apparatus. The storage medium may be a machine-readable signal medium or a machine-readable storage medium. Alternatively, the storage medium may be a non-temporary computer-readable storage medium, for example, a non-temporary computer-readable storage medium may be a ROM, a random access memory (Random Access Memory, RAM), a compact disc read-only memory (Compact Disc ROM, CD-ROM), a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0244] An embodiment of the present invention provides a computer program product, including a computer program / instruction, which implements the above-mentioned data partition storage method when executed by a processor.

[0245] The above description is only the principle and preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this field, several other modifications can be made based on the principle of the present invention, which should also be considered as the protection scope of the present invention.

Claims

1. A method for data partition storage, characterized in that: include: A data acquisition step, for acquiring traffic node data of a plurality of traffic nodes, wherein the traffic node data at least includes traffic node information of the traffic nodes and connection relationships between the traffic nodes, and the traffic node information at least includes traffic flow data; A similarity matrix construction step is used to construct a structural similarity matrix according to the connection relationship between the traffic nodes, to construct a flow similarity matrix according to the traffic flow data in the traffic node information, and to construct a joint similarity matrix according to the structural similarity matrix and the flow similarity matrix; The spectral clustering division step is used to divide the plurality of traffic nodes into a plurality of sub-clusters according to the joint similarity matrix, and store the traffic node data corresponding to different sub-clusters into different storage areas.

2. The method for data partition storage according to claim 1, characterized in that: The step of constructing a structural similarity matrix according to the connection relationship between the traffic nodes includes: An adjacency matrix between traffic nodes is constructed, and the adjacency matrix is ​​used as a structural similarity matrix. Elements in the adjacency matrix represent whether two traffic nodes are connected.

3. The method for data partition storage according to claim 1, characterized in that: The constructing of a traffic similarity matrix according to the traffic flow data in the traffic node information includes: The traffic similarity matrix is ​​constructed as: Among them, S f is the traffic similarity matrix, DTW is the similarity function, is the standard value of the traffic flow data of the ith traffic node, is the standard value of the traffic flow data of the jth traffic node.

4. The method for data partition storage according to claim 1, characterized in that: The constructing a joint similarity matrix according to the structural similarity matrix and the traffic similarity matrix comprises: The joint similarity matrix is ​​constructed as: S=αS s +βS f , Among them, S is the joint similarity matrix, α is the first weight coefficient, β is the second weight coefficient, and α+β=1, S s is the structural similarity matrix, S f is the traffic similarity matrix.

5. The method for data partition storage according to claim 1, characterized in that: The dividing the plurality of traffic nodes into a plurality of sub-clusters according to the joint similarity matrix comprises: Calculating a joint Laplacian matrix according to the joint similarity matrix; Solving the eigenvectors corresponding to the first K smallest eigenvalues ​​of the Laplace matrix, and forming a characteristic matrix with the eigenvectors corresponding to the first K smallest eigenvalues; Each row of the feature matrix corresponds to a feature vector of the traffic node, and each column of the feature matrix represents a feature of the traffic node. Clustering calculation is performed on the feature matrix to cluster the multiple traffic nodes into different sub-clusters.

6. The method for data partition storage according to claim 1, characterized in that: The spectral clustering division step also includes a dynamic adjustment step, which includes: executing the data acquisition step, the similarity matrix construction step and the spectral clustering division step once every preset time period.

7. The method for storing data in partitions according to any one of claims 1 to 6, characterized in that: The method further comprises an access verification step: when the storage area is accessed, the access rights and / or access rules of the visitor are verified, and the access rights and / or access rules of different storage areas are different.

8. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to at least one of the processors; wherein, The memory stores instructions that can be executed by at least one of the processors, and the instructions are executed by at least one of the processors to enable at least one of the processors to execute the method for partitioning data storage as described in any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium stores computer instructions, and when a computer executes the computer instructions, it is used to execute the data partition storage method according to any one of claims 1 to 7.

10. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instruction is executed by a processor, the method for partitioning data storage as claimed in any one of claims 1 to 7 is implemented.

Images