Vehicle trajectory flow data space-time hybrid index and query method and system and storage medium

Abstract

The invention discloses a time-space hybrid indexing and querying method and system for vehicle trajectory flow data and a storage medium, and belongs to the field of information retrieval. The method comprises the following steps: collecting vehicle trajectory flow data, and performing spatio-temporal expression and modeling on the vehicle trajectory flow data; performing time division on the vehicle trajectory flow data to obtain time partitions; performing space division on the trajectory stream data of each time partition to obtain a division result; based on a division result, constructing a space-time T-SP-TL learning index; on the basis of newly input data, the constructed T-SP-TL learning index is maintained and updated; and completing time-space integrated query based on the T-SP-TL learning index. The method can be applied to space-time query of vehicle trajectory flow big data, organization and dynamic management of geographic big data can be completed under the efficient and accurate condition, the trajectory flow data access efficiency and real-time updating performance are improved, and auxiliary support is provided for effective traffic control and dispersion, public safety monitoring and early warning, emergency rescue and other applications.

Description

Technical Field

[0001] The present invention belongs to the field of information retrieval technology, and in particular relates to a spatiotemporal hybrid indexing and query method, system and storage medium for vehicle trajectory flow data. Background Art

[0002] The era of spatiotemporal big data has driven the shift of spatiotemporal intelligence toward a "data-intensive paradigm," but this has also led to increasingly prominent challenges: "massive data, information explosion, and limited knowledge." Efficient and effective access is fundamental to the efficient transformation of spatiotemporal "big data" into "big knowledge," and spatiotemporal indexing is key to improving access efficiency. Due to the rapid growth in data volumes, existing indexing methods are struggling to meet current access needs. To address these challenges with existing indexing structures, researchers have applied machine learning methods to indexing and proposed the concept of learned indexing. The key idea is to replace traditional indexing with learning, converting queries into model predictions and leveraging the computational power of computers to improve access efficiency.

[0003] Trajectory stream data is the sequence of positions generated by the movement of mobile objects in geographic space. It contains not only spatial information about each location point at each moment, but also attribute information about the stream sequence itself. As the volume of data continues to increase over time, it is a core component of geographic big data. Trajectory stream data contains rich, time-sensitive information. Real-time analysis, immediate discovery, and rapid response are essential for effectively leveraging the value of trajectory stream data (for example, collision detection for self-driving cars). Compared to other types of big data, trajectory stream data boasts real-time data arrival, rapid growth, and large data volumes. Considering the rapid growth of trajectory stream data, the importance of indexing methods becomes even more pronounced. Existing spatiotemporal indexing methods effectively meet the spatiotemporal query requirements of historical data, but they overlook the dynamic spatiotemporal distribution of trajectory streams, making it difficult to support real-time updates and efficient spatiotemporal query processing of trajectory stream data.

[0004] A learned index treats the index itself as a learning model. By training and learning the relationship between data and physical locations, it derives an adaptive mapping function f(key)->pos. Given a key to be queried (key), it returns the position (pos) of the record containing that key. Existing tree-based indexes perform queries through a series of comparisons, while learned indexes use direct computation, thus reducing data access time and conserving memory resources. Learned indexes require relatively static and one-dimensionally ordered data, but trajectory stream data is not naturally ordered and continuously and dynamically increases, making them impractical to apply directly to trajectory stream data.

[0005] Therefore, existing indexing methods find it difficult to effectively process trajectory stream data because their access efficiency drops sharply when the new trajectory data stream exceeds the capacity of the spatiotemporal index extension; trajectory stream data is continuously updated and has a huge scale, causing the index structure to change dramatically. Summary of the Invention

[0006] This invention aims to address the shortcomings of existing technologies by proposing a spatiotemporal hybrid indexing and querying method, system, and storage medium for vehicle trajectory data. By integrating innovative ideas and methods such as big data, graph network analysis, and learning indexing, this method addresses the need for efficient trajectory data management and mathematically quantifies the spatiotemporal proximity, spatiotemporal heterogeneity, spatiotemporal similarity, dynamic processes, and interactions of trajectory data within a road network space. It also establishes a spatiotemporal adaptive partitioning method for trajectory data and constructs a spatiotemporal learning index for multi-objective tasks. It also implements a dynamic update and maintenance algorithm and a spatiotemporal access algorithm that adapts to multiple levels and multiple objectives. Based on learning indexing theory and methods, the retrieval problem is transformed into a prediction problem, constructing a spatiotemporal learning indexing method that self-balances access and update efficiency. This new engine supports high-efficiency geoscientific applications such as real-time trajectory data mining, scheduling, and geographic knowledge discovery within road network environments.

[0007] To achieve the above objectives, the present invention provides the following solution: a spatiotemporal hybrid indexing and query method for vehicle trajectory stream data, comprising the following steps:

[0008] S1. Collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data;

[0009] S2. Divide the vehicle trajectory flow data into time partitions to obtain time partitions; and divide the trajectory flow data of each time partition into space to obtain a partition result;

[0010] S3. Based on the division result, construct a spatiotemporal T-SP-TL learning index;

[0011] S4. Maintain and update the constructed T-SP-TL learning index based on the newly input data;

[0012] S5. Complete spatiotemporal integrated query based on the T-SP-TL learning index.

[0013] Further preferably, S1 comprises the following steps:

[0014] S11. Use mathematical representation methods to express the road network and uniquely code each road segment;

[0015] S12. Determine the correlation between road sections based on the road section topology relationship information, and establish a comprehensive correlation model;

[0016] S13, analyzing the local similarity of the trajectory flow data and obtaining a local similarity measurement index;

[0017] S14. Based on the comprehensive correlation model, spatiotemporal correlation and heterogeneity indicators, a spatiotemporal comprehensive similarity measurement model is constructed.

[0018] Further preferably, in S2, the method for obtaining the time partition includes:

[0019] S21. Based on the time series distribution curve and data volume of trajectory flow data, combined with time series analysis methods, solve the time division nodes;

[0020] S22. Based on the time partition node, a time partition table is constructed to obtain the time partition; the time partition table includes a time period attribute and a pointer.

[0021] Further preferably, the method for obtaining the division result includes:

[0022] S23, setting a termination condition based on the number of trajectory point objects;

[0023] S24, constructing a linear graph based on the road network graph and the trajectory flow object set;

[0024] S25 , decomposing the linear graph into several subgraphs, and adding each subgraph to the hierarchical partitioning result set until a termination condition is met, otherwise looping S25 .

[0025] Further preferably, S3 comprises the following steps:

[0026] S31. Divide the time partition into a historical time partition and a current time partition according to the characteristics of the time partition;

[0027] S32: Construct a T-SP-TL learning index for the historical time partition and the current time partition.

[0028] Further preferably, the method for constructing an index for the historical time partition includes:

[0029] S321, converting the linear graph into a road network graph, with the sub-network of the road network graph serving as a child node of the root node;

[0030] S322. Identify bridge points for each sub-network and add them to the corresponding tree nodes; and calculate the distance matrix from the leaf nodes to the root node;

[0031] S323. Encode each trajectory object in the trajectory dataset of the leaf node, and divide the trajectory dataset of the leaf node into several subsets according to the time attribute; generate a second-layer model for each subset; use a neural network model for learning, where the input of the model is the key of the data record and the training label is the position of the key;

[0032] S324. After the second-layer model is constructed, the first-layer model is constructed; the query key of the first-layer model is the independent variable, and the dependent variable is the serial number of the second-layer model, completing the index construction of the historical time partition.

[0033] Further preferably, the method for constructing an index for the current time partition includes:

[0034] S325, performing spatial distribution feature analysis on the historical trajectory flow data, and performing the spatial division step in S2 based on the analysis result to obtain the division result;

[0035] S326: Based on the division result, execute step S321;

[0036] S327. Perform similarity measurement on the results of the spatial distribution feature analysis and select the spatial division area that is most similar to the historical time division;

[0037] S328: Use the second layer model corresponding to the spatial partition area selected in S327 as the initial network, and construct a two-layer learning index based on the initial network and the trajectory flow data of the current time partition.

[0038] Further preferably, S4 comprises the following steps:

[0039] S41. Connect a buffer page to the last second-layer model in each time partition. When data is updated, save the updated data in the buffer page until the buffer page overflows. When the buffer page overflows, directly input the updated data into the second-layer model to retrain the second-layer model.

[0040] S42: When the distribution of the updated data changes and the error of the second-layer model is greater than a preset error, the updated data is split into two and a second-layer model is added; the second-layer model is used to learn the split data;

[0041] S43: Optimize the parameters of the first layer model to complete the maintenance and update of the T-SP-TL learning index.

[0042] The present invention also provides a spatiotemporal hybrid indexing and query system for vehicle trajectory flow data, the system being used to implement the method described above, comprising: a modeling unit, a partitioning unit, a construction unit, an updating unit, and an application unit;

[0043] The modeling unit is used to collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data;

[0044] The division unit is used to divide the vehicle trajectory flow data into time partitions to obtain time partitions; and to divide the trajectory flow data of each time partition into space to obtain a division result;

[0045] The construction unit is used to construct a spatiotemporal T-SP-TL learning index based on the division result;

[0046] The updating unit is used to maintain and update the constructed T-SP-TL learning index based on newly input data;

[0047] The application unit is used to complete the spatiotemporal integrated query based on the T-SP-TL learning index.

[0048] The present invention also provides a vehicle trajectory flow data spatiotemporal hybrid index and query storage medium, wherein the storage medium stores a computer program, and the computer program is executed by a processor to implement the above method.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] This invention, based on a multi-objective-driven adaptive spatiotemporal data partitioning method, addresses the massive and dynamic nature of trajectory stream data by designing a highly adaptable spatiotemporal learning index structure. This structure also constructs an integrated spatiotemporal access algorithm and establishes an update and maintenance method that integrates internal and external memory with global-local collaboration. This reduces the number of I / O operations and the frequent internal and external memory interactions of index nodes, thereby improving the efficiency and timeliness of real-time updates and spatiotemporal queries for trajectory stream data. This method has been successfully applied to the dynamic management of trajectory stream data, achieving accurate and efficient results while maintaining effectiveness, demonstrating outstanding technical advantages. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solution of the present invention, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 Schematic diagram of the flow of a spatiotemporal hybrid indexing and query method for vehicle trajectory stream data according to an embodiment of the present invention;

[0053] Figure 2 This is a schematic diagram of the main process of converting a road network graph into a linear graph partitioning according to an embodiment of the present invention;

[0054] Figure 3 This is a schematic diagram of a road network graph division result according to an embodiment of the present invention;

[0055] Figure 4This is a diagram of the T-SP-TL learning index structure framework according to an embodiment of the present invention;

[0056] Figure 5 Schematic diagram of TL learning index according to an embodiment of the present invention. DETAILED DESCRIPTION

[0057] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only 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 making creative efforts are within the scope of protection of the present invention.

[0058] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0059] Example 1:

[0060] like Figure 1 As shown, this embodiment provides a spatiotemporal hybrid indexing and query method for vehicle trajectory stream data, including the following steps:

[0061] S1. Collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data.

[0062] Specifically, the steps include:

[0063] S11. Use mathematical representation methods of complex networks to digitally express the road network and uniquely encode each road section.

[0064] in,

[0065] Road network graph: The real road network can be represented as G = (V, E), where V is a series of road vertices, V = {v1, v2, v3, ..., v n}; E represents a road segment and connects the road vertices, E = {e1, e2, e3, ..., e n}.

[0066] Trajectory: Tra=(moid,[pos1,pos2,...,pos n ],[t1,t2...,t n ]); is a triple, where moid is the unique identifier of the trajectory object, pos n is the location of the trajectory point object (including the identifier of the road segment where the object is located), t n is the trajectory point pos n time.

[0067] Linear graph: Given a road network graph G = (V, E), its linear graph can be expressed as The construction method is: convert the road segments in G into If the road segments in G are related to each other, then construct Each vertex has a weight, which is the number of spatial objects on the road segment.

[0068] Bridge point: Given a road network graph G =<V,E> , its subgraph is subG ik and subG ij , if a vertex v is in subG ik and subG ij In the case of vertex v, it is called a bridge point, subG ij The bridge point set is B(subG ij )express,

[0069] Joint bridge point: given a subnetwork subG of G ij , the set of bridge points of all its sub-networks is called the joint bridge point, where each element in the joint bridge point is unique. Therefore, the joint bridge point set of a non-leaf node is the union of all bridge points of its child nodes, and the joint bridge point set of a leaf node is equal to the set of its bridge points. The set of joint bridge points is denoted as UB(subG ij ).

[0070] Distance Matrix: Each tree node has a distance matrix. The rows and columns of the distance matrix for non-leaf nodes represent the joint bridge points of non-leaf nodes, and each value is the shortest path distance between two vertices. In the distance matrix for leaf nodes, the rows represent bridge points, the columns represent vertices in the leaf node, and each value represents the shortest path distance between the bridge point and its vertex.

[0071] Valid partition / partially valid partition / invalid partition: given a subnetwork subG i , a query point q, range distance d r If the distance between the partition's bridge point and the query point is less than d r (SPDis(bi,q)≤d r ), subG i It is called a valid partition; if the distance between the bridge point and the query point is greater than d r (SPDis(bi,q)>d r ), is called an ineffective partition. In particular, if its partition contains the query point, it is called a partially effective partition; if the distance between the bridge point and the query point is not less than d r, which is called a partially valid partition.

[0072] S12. Based on the topological relationship information of the road sections, combined with indicators such as the road section grade and flow ratio, the correlation between the road sections is determined, and a comprehensive correlation model is established.

[0073] The association model RM = (N, J, S) represents the association between road segments, where N represents the road segments in the network; J indicates whether the corresponding road segments are directly adjacent (i.e., they share a common endpoint); and the association set S represents the degree of association between adjacent road segments. For each road segment, abstracted as a node, the weight set S is further defined by representing the traffic conditions of the road segment (such as average flow rate). The dimension of S is equal to the number of associated road segments.

[0074] S13. Analyze the local similarity of the trajectory data and obtain a local similarity measurement index. In the specific example, the Euclidean distance and Pearson correlation coefficient are used to measure the local similarity. The similarity index matrix ST is obtained by weighted combination of the Pearson correlation coefficient and the Euclidean distance:

[0075] ST=ω1D+ω2P,

[0076] Where ω1 and ω2 are weight coefficients; D is the Euclidean distance matrix, and P is the Pearson correlation coefficient matrix.

[0077] S14. Based on the comprehensive correlation model, spatiotemporal correlation and heterogeneity index, a spatiotemporal comprehensive similarity measurement model is constructed to further realize the analysis of spatiotemporal comprehensive similarity. In a specific example, the similarity between road segments is defined by combining the road network correlation model RM, the similarity index matrix ST and the Gaussian similarity function:

[0078]

[0079] Where μ is the parameter of Gaussian similarity function; st ij Represents the data in the i-th row and j-th column of the similarity index matrix.

[0080] S2. Perform time division on the vehicle trajectory flow data to obtain time partitions; and perform spatial division on the trajectory flow data of each time partition to obtain a division result.

[0081] The methods for obtaining time partitions include:

[0082] S21. Based on the time series distribution curve and data volume, combined with the time series analysis method, solve the time division node.

[0083] S22. Based on the time partition nodes, a time partition table (TP) is constructed to obtain time partitions; the time partition table records the index of the corresponding time interval, which includes a time period attribute and a pointer to a sub-index.

[0084] The method for obtaining the division result includes:

[0085] S23. Setting a termination condition based on the number of trajectory point objects.

[0086] For a given road network graph G and trajectory object set Trans, the number of trajectory point objects contained in the leaf partitions does not exceed β as the termination condition.

[0087] S24. Construct a linear graph based on the road network graph G and the trajectory flow object set Trans like Figure 2 As shown in Figure 1, it is the main process of converting a road network graph into a linear graph.

[0088] S25. Linear graph Decompose into several subgraphs with equal weights Each subgraph is added to a set of spatial partitioning results until the termination condition is met. The partitioning results are as follows: Figure 3 As shown. That is, when any bottom-level subgraph in the hierarchical partitioning result set contains more than β objects, otherwise loop S25. During the hierarchical partitioning process, the partitioning criterion is to minimize the number of associated edges of the edge cut while ensuring a relatively balanced number of objects in each partition space.

[0089] S3. Based on the partitioning results, construct a spatiotemporal T-SP-TL learning index.

[0090] In this embodiment, the learning index has the same meaning as the second-layer model and the first-layer model described below. In the specific implementation, different terms are used in different stages. Specifically, S3 includes the following steps:

[0091] S31. Divide the time partition into a historical time partition and a current time partition according to the characteristics of the time partition.

[0092] S32: Construct T-SP-TL learning index for historical time partition and current time partition. Figure 4 The figure shows the framework diagram of the T-SP-TL learning index structure.

[0093] Methods for building indexes for historical time partitions include:

[0094] S321, the sub-linear graph Convert to subnet graph subG i, the root node corresponds to the entire road network graph G, which is a single leaf node and initially does not contain any other tree nodes; the sub-networks of the road network graph serve as child nodes of the root node, the root node becomes a non-leaf node, and each sub-network corresponds to an SP tree node. The relationship between tree nodes is equal to the relationship between sub-networks, and a sub-network without sub-networks is a leaf node.

[0095] S322: Identify the bridge point for each sub-network and add it to the corresponding tree node.

[0096] S323, encode each trajectory object in the trajectory dataset of the leaf node, and the encoding rule is the time feature of the trajectory point year-month-day-hour-minute-second; divide the trajectory dataset TraD of the leaf node into n subsets {TraD1, TraD2, ..., TraD n} (e.g., the time granularity is set to 1h); the data regions are disjoint and ordered in size, i.e., the time attribute of the data in the i+1th data region is larger than the time attribute in the i-th data region; each subset generates a second-layer model f i (x); For each subset, a neural network model is used for learning, where the input of the model is the key (encoded value) of the data record and the training label is the position of the key.

[0097] To ensure the accuracy of the trained second-layer model's predictions, a threshold is set for the average error of the second-layer model, denoted as Maxerror. When the error exceeds Maxerror, the data region corresponding to the second-layer model is split again, dividing the data region into two. At the same time, a second-layer model is added. At this time, the number of data set partitions becomes n+1, and the number of second-layer models is also increased by 1. The split data subsets are relearned until the prediction error is less than Maxerror.

[0098] S324. After the second-layer model is constructed, the first-layer model F(x) is constructed. The query key (coded value) of the first-layer model is the independent variable, and the dependent variable is the number of the corresponding second-layer model. The index construction of the historical time partition is completed. The schematic diagram of the TL learning index (two-layer learning index) structure is as follows: Figure 5 As shown, model 0 is the first layer model, and models 1 to K are the second layer models.

[0099] Methods for building indexes for the current time partition include:

[0100] In the case of continuous trajectory stream data, since the spatial distribution characteristics of the data within the current time interval are unknown, to reduce the frequent adjustment of the spatial partitioning of the current dataset, the spatial distribution characteristics of the data within the current time interval can be detected and predicted based on the historical dataset. The spatial distribution characteristics of the historical trajectory stream data are analyzed, and based on the analysis results, the spatial partitioning step in S2 is performed to obtain a partitioning result.

[0101] S326. Based on the division result, execute step S321.

[0102] S327: Measure the similarity between the spatial distribution characteristics of the current time partition data and the historical time partition data, and select the spatial partition area that is most similar to the historical time partition;

[0103] S328: Use the second layer model corresponding to the spatial partition area selected in S327 as the initial network to avoid frequent training of the neural network model. Construct a two-layer learning index based on the initial network and the trajectory flow data of the current time partition.

[0104] S4. Maintain and update the constructed T-SP-TL learning index based on the newly input data.

[0105] Specifically, S4 includes the following steps:

[0106] S41. Index update mainly involves the current index structure, and the update usually only involves the learning index part. In order to solve the problem of large-scale data movement caused by the insertion of new data, a buffer page is connected to the last second-level model in each time partition to store the newly inserted data.<key,pos> Yes, key refers to the encoding value of the trajectory object, and pos is the location of the data. When the data is updated, the updated data is saved in the buffer page until the buffer page overflows. When the buffer page overflows, the updated data is directly input into the corresponding second-layer model, and the second-layer model is retrained to fit the new data distribution, thereby avoiding frequent training of the neural network.

[0107] S42. When the distribution of updated data changes and the amount of data that each second-layer model is responsible for increases, resulting in the model being unable to learn the distribution of data well, in order to ensure that the retrained model predictions are as accurate as possible, when the error of the retrained model is greater than Maxerror, the data area corresponding to the model is split; the updated data is divided into two, and a second-layer model is added at the same time; at this time, the number of data set partitions becomes n+1, and the number of second-layer models is also increased by one, and the second-layer model is used to learn the split data.

[0108] S43. Optimize and adjust the parameters of the first-layer model to complete the maintenance and update of the T-SP-TL learning index.

[0109] S5. Complete spatiotemporal integrated query based on T-SP-TL learning index.

[0110] The space-time integrated query is: given a space-time range query Q = (Q s , Q t ), where Q s =(q, d r ), Q t =[t s , t e ], Q represents the query of all valid objects whose shortest path distance to q is no greater than d r , and the object's time is given in Q t within the range.

[0111] Calculate Q t The intersection of the time intervals in the TP table gives a set of time intervals Q 1t , Q 2t ,…,Q kt ; The initial query is mapped to a set of k queries, where Q i =(Q s , Q it ), i=1,...,k, each query Q i By using the corresponding spatiotemporal query method. Calculate the shortest path distance between the query point q and the bridge point of the leaf node leaf(q) containing the query point q. If leaf(q) is identified as a non-valid leaf node, leaf(q) will be inserted into the valid tree node set (EN). If leaf(q) is a partially valid node, execute the neighboring partition expansion method, which can search for tree nodes that may contain valid data within the query range, and add the valid or partially valid nodes to the valid node set EN; for the leaf partitions that may be valid in the set EN, according to Q t , calculate the corresponding encoding value key1, key2, ..., key n , determine the second-layer model number through the first-layer model; predict the storage location corresponding to the data record that meets the conditions from the corresponding second-layer model, so that the data that meets the conditions can be found and input into P obj In the data group, find the object P that meets the conditions obj Refine and identify data objects that meet the query conditions.

[0112] Example 2:

[0113] This embodiment provides a spatiotemporal hybrid indexing and query system for vehicle trajectory stream data, including: a modeling unit, a partitioning unit, a construction unit, an updating unit, and an application unit.

[0114] The modeling unit is used to collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data.

[0115] The division unit is used to divide the vehicle trajectory flow data into time partitions to obtain time partitions; and to divide the trajectory flow data of each time partition into space to obtain division results.

[0116] The construction unit is used to construct a spatiotemporal T-SP-TL learning index based on the partitioning results.

[0117] The updating unit is used to maintain and update the constructed T-SP-TL learning index based on the newly input data.

[0118] The application unit is used to complete spatiotemporal integrated query based on the T-SP-TL learning index.

[0119] A vehicle trajectory flow data spatiotemporal hybrid indexing and query storage medium is also provided, wherein the storage medium stores a computer program, and the computer program is executed by a processor to implement the above method.

[0120] Example 3:

[0121] The query performance of the T-SP-TL learned index was tested on both real trajectory data sets and artificially generated data sets. For the synthetic data set, the Brinkhoff data generator was used to generate trajectory data sets based on the urban road networks of Oldenburg (Germany) and San Joaquin (USA). The Oldenburg network contains 6,105 road vertices and 7,034 road segments. The San Joaquin network is approximately three times the size of the Oldenburg network, with 18,496 road vertices and 24,123 road segments. The performance of spatiotemporal range queries was evaluated. For the real data set, urban taxi trajectory data sets from Nanjing and Suzhou, China were used. The Nanjing trajectory data set contains approximately 480,000 items, while the Suzhou trajectory data set contains approximately 1.02 million items. Index update and query performance were evaluated.

[0122] For query performance, we selected a spatial range of 1% of the city's area and a temporal range of 1% and performed 20 random queries to evaluate query time. The experimental results showed that the query time for the Oldenburg dataset was 10.34ms; for the San Joaquin dataset, the query time was 11.63ms; for the Suzhou dataset, the query time was 66.87ms; and for the Nanjing dataset, the query time was 44.47ms. For update performance, we evaluated data updates within a single temporal partition. The update time for the Oldenburg dataset was 402.07s; for the San Joaquin dataset, the update time was 807.79s; for the Nanjing dataset, the update time was 996.04s; and for the Suzhou dataset, the update time was 2034.13s.

[0123] The embodiments described above are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Without departing from the spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by persons skilled in the art should fall within the scope of protection defined by the claims of the present invention.

Claims

1. A spatiotemporal hybrid indexing and query method for vehicle trajectory stream data, characterized in that: The following steps are involved: S1. Collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data; S2. Divide the vehicle trajectory flow data into time partitions to obtain time partitions; and divide the trajectory flow data of each time partition into space to obtain a partition result; S3. Based on the division result, construct a spatiotemporal T-SP-TL learning index; S4. Maintain and update the constructed T-SP-TL learning index based on the newly input data; S5. Complete spatiotemporal integrated query based on the T-SP-TL learning index.

2. The spatiotemporal hybrid indexing and querying method for vehicle trajectory stream data according to claim 1 is characterized in that: S1 includes the following steps: S11. Use mathematical representation methods to express the road network and uniquely code each road segment; S12. Determine the correlation between road sections based on the road section topology relationship information, and establish a comprehensive correlation model; S13, analyzing the local similarity of the trajectory flow data and obtaining a local similarity measurement index; S14. Based on the comprehensive correlation model, spatiotemporal correlation and heterogeneity indicators, a spatiotemporal comprehensive similarity measurement model is constructed.

3. The spatiotemporal hybrid indexing and querying method for vehicle trajectory stream data according to claim 1 is characterized in that: In S2, the method for obtaining the time partition includes: S21. Based on the time series distribution curve and data volume of trajectory flow data, combined with time series analysis methods, solve the time division nodes; S22. Based on the time partition node, a time partition table is constructed to obtain the time partition; the time partition table includes a time period attribute and a pointer.

4. The spatiotemporal hybrid indexing and querying method for vehicle trajectory stream data according to claim 1 is characterized in that: The method for obtaining the division result includes: S23, setting a termination condition based on the number of trajectory point objects; S24, constructing a linear graph based on the road network graph and the trajectory flow object set; S25 , decomposing the linear graph into several subgraphs, and adding each subgraph to the hierarchical partitioning result set until a termination condition is met, otherwise looping S25 .

5. A spatiotemporal hybrid indexing and query method for vehicle trajectory stream data according to claim 4, characterized in that: S3 includes the following steps: S31. Divide the time partition into a historical time partition and a current time partition according to the characteristics of the time partition; S32: Construct a T-SP-TL learning index for the historical time partition and the current time partition.

6. A spatiotemporal hybrid indexing and query method for vehicle trajectory stream data according to claim 5, characterized in that: The method for constructing an index for the historical time partition includes: S321, converting the linear graph into a road network graph, with the sub-network of the road network graph serving as a child node of the root node; S322. Identify the bridge point for each sub-network and add it to the corresponding tree node; S323. Encode each trajectory object in the trajectory dataset of the leaf node, and divide the trajectory dataset of the leaf node into several subsets according to the time attribute; generate a second-layer model for each subset; use a neural network model for learning, where the input of the model is the key of the data record and the training label is the position of the key; S324. After the second-layer model is constructed, the first-layer model is constructed; the query key of the first-layer model is the independent variable, and the dependent variable is the serial number of the second-layer model, completing the index construction of the historical time partition.

7. The spatiotemporal hybrid indexing and querying method for vehicle trajectory stream data according to claim 5, characterized in that: The method for constructing an index for the current time partition includes: S325, performing spatial distribution feature analysis on the historical trajectory flow data, and performing the spatial division step in S2 based on the analysis result to obtain the division result; S326: Based on the division result, execute step S321; S327. Perform similarity measurement on the results of the spatial distribution feature analysis and select the spatial division area that is most similar to the historical time division; S328: Use the second layer model corresponding to the spatial partition area selected in S327 as the initial network, and construct a two-layer learning index based on the initial network and the trajectory flow data of the current time partition.

8. The spatiotemporal hybrid indexing and querying method for vehicle trajectory stream data according to claim 6 is characterized in that: S4 includes the following steps: S41. Connect a buffer page to the last second-layer model in each time partition. When data is updated, save the updated data in the buffer page until the buffer page overflows. When the buffer page overflows, directly input the updated data into the second-layer model to retrain the second-layer model. S42: When the distribution of the updated data changes and the error of the second-layer model is greater than a preset error, the updated data is split into two and a second-layer model is added; the second-layer model is used to learn the split data; S43: Optimize the parameters of the first layer model to complete the maintenance and update of the T-SP-TL learning index.

9. A spatiotemporal hybrid indexing and query system for vehicle trajectory stream data, the system being used to implement the method according to any one of claims 1 to 8, characterized in that: include: Modeling unit, partitioning unit, construction unit, update unit and application unit; The modeling unit is used to collect vehicle trajectory flow data and perform spatiotemporal expression and modeling on the vehicle trajectory flow data; The division unit is used to divide the vehicle trajectory flow data into time partitions to obtain time partitions; and to divide the trajectory flow data of each time partition into space to obtain a division result; The construction unit is used to learn the spatiotemporal T-SP-TL index based on the division result; The updating unit is used to maintain and update the constructed T-SP-TL learning index based on newly input data; The application unit is used to complete the spatiotemporal integrated query based on the T-SP-TL learning index.

10. A vehicle trajectory flow data spatiotemporal hybrid index and query storage medium, characterized in that: The storage medium stores a computer program, and the computer program is executed by a processor to implement the method according to any one of claims 1 to 8.

Images