Truck travel category division method and system based on multi-scale hypergraph convolution depth clustering

Abstract

The invention relates to the field of traffic transportation information engineering, and discloses a truck travel category division method and system based on multi-scale hypergraph convolution depth clustering, and the method comprises the steps: constructing a truck travel feature matrix under each spatial scale; constructing a truck travel mode hypergraph under each spatial scale, and modeling a many-to-many high-order association relationship of trucks in a plurality of spatial scales; a multi-scale hypergraph convolution depth clustering model is constructed, and training and testing of truck travel category division are carried out; and carrying out instance verification of truck travel category division. The model is applied to a series of empirical data sets and self-collection data sets, and it is proved that the model can obtain good division performance superior to a baseline model.

Description

Technical Field

[0001] The present invention relates to the field of transportation information engineering technology, and in particular to a method and system for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering. Background Art

[0002] Accurate classification of groups is an important basis for in-depth understanding of group differences, classification and formulation of policy opinions, and precise initiation of emergency response. When completing cargo transportation operations, trucks constantly travel across multi-scale spaces such as public transportation places, streets, and administrative areas, and thus there are "many-to-many" associations at various scales and travel behaviors that interact across spatial scales. Therefore, when classifying truck travel modes, it is necessary to fully consider the "many-to-many" high-order association characteristics of travel modes between trucks that interact at multiple spatial scales.

[0003] Existing studies have used classic algorithms such as k-means, DBSCAN, and hierarchical clustering to classify truck travel modes, but have not used the personalized travel association characteristics between trucks to influence the clustering results. In addition, related technologies also use graph clustering algorithms to learn the implicit feature representations behind the original data and capture the implicit association structure between individuals, providing a breakthrough for group classification. However, the traditional graph method of "connecting two points on a single edge" is not sufficient to express the "many-to-many" mapping attribute between vertices, which to a certain extent hinders the performance of group classification, and will lead to inaccurate group classification of trucks. Summary of the invention

[0004] The present invention provides a method and system for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering, which is used to solve the defect of inaccurate group classification of trucks in related technologies. In the solution of the present application, the travel records of trucks can be clustered by constructing a hypergraph, thereby improving the accuracy of group classification.

[0005] The present invention provides a method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering, comprising:

[0006] Step 1: Construct the travel feature matrix of trucks at each spatial scale. According to the historical travel record set of N trucks, obtain the historical travel record set of N trucks in a given period. At the spatial scale s, extract the F-dimensional travel features for each truck and construct the travel feature matrix of N trucks. The truck travel feature matrix X = {X1,…,X s ,…,X S}, where x s,i is the travel characteristic expression of the i-th truck at spatial scale s, 1≤i≤N; x iis the travel characteristic expression of the i-th truck, 1≤i≤N; the F-dimensional travel characteristics of each truck come from the time, space and attribute dimensions respectively. The time dimension includes the number of truck driver fatigue driving, night driving time, truck departure time and average daily travel time; the space dimension includes the truck travel distance, truck stop function and truck travel average speed; the attribute dimension includes license plate color and transportation trip number;

[0007] Step 2: construct a hypergraph of truck travel patterns at each spatial scale, model the many-to-many high-order association relationship of trucks, and calculate the truck travel feature matrix X according to step 1. s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel mode s ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s ;H s is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighboring groups, H s The internal element h(v,e) represents the affiliation relationship between node v and hyperedge e. The specific calculation method is shown in formula (1):

[0008]

[0009] Among them, v∈e means that the node is on the hyperedge;

[0010] Step 3: Build a multi-scale hypergraph convolutional deep clustering model to train and test the truck travel classification, including:

[0011] Step 3.1, using the autoencoder model, learn the implicit travel mode representation of trucks;

[0012] Step 3.2, using the hypergraph convolutional network structure, learn the implicit travel mode representation of the truck neighborhood group in the local space;

[0013] Step 3.3, using the cross self-attention mechanism, the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales are integrated, and deep clustering is performed through the hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales;

[0014] Step 3.4, training and testing the truck travel classification performance;

[0015] Step 4: Example verification, input the travel feature matrix X of N trucks at the S+1th spatial scale S+1 , construct a truck travel pattern hypergraph G′ that integrates multi-scale spatial features S+1 , realize the classification of truck travel groups; applied to the ACM dataset, Cite dataset and truck GPS dataset, through comparison and ablation experiments, verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering.

[0016] According to the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by the present invention, the method of using an autoencoder model to learn the implicit travel mode representation of trucks includes:

[0017] The autoencoder model is introduced to learn the truck travel feature matrix X at the spatial scale s in step 1 through the built-in encoder. s Implicit travel mode representation at convolutional layer l (1≤l≤L) Calculated by formula (2):

[0018]

[0019] In the formula, σ ReLU is a linear activation function; AutoEncoder is an autoencoder. When l=1, The truck travel feature matrix X output in step 1 s ; W (l) is the convolution kernel weight of the lth layer, b (l) is the bias value of the lth layer, both of which are weight parameters of the model to be trained;

[0020] Through the decoder built into the autoencoder, after L convolutional layers, the reconstructed Complete the decoding operation; minimize the average reconstruction error produced by the autoencoder model at all spatial scales Formula (3) is used to express the truck travel mode that is closest to the actual situation:

[0021]

[0022] Among them, ||·|| Fro is the F norm;

[0023] For the spatial scale s, according to the truck travel characteristic matrix X s The distribution of truck travel categories in real scenarios is obtained by obtaining the probability distribution curve GT; according to the implicit feature representation of truck travel mode in the Lth convolutional layer The distribution of the truck travel category probability distribution curve TR obtained by the encoder division is obtained, and the cumulative error of the measurement curve GT and curve TR at all spatial scales is calculated. Calculated by formula (4):

[0024]

[0025] Among them, GT sij and AE sij are the probabilities that truck i belongs to category j in curve GT and curve AE at spatial scale s, respectively.

[0026] According to the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by the present invention, the hypergraph convolution network structure is used to learn the implicit travel mode representation of the truck neighborhood group in the local space, including:

[0027] According to the truck travel mode hypergraph G at any spatial scale s obtained in step 2 s , perform hypergraph convolutional network operations to capture the local spatial topological changes of the truck neighborhood group. For the lth convolutional layer, the implicit travel mode of the truck neighborhood group Calculated by formula (5):

[0028]

[0029] Where Θ is the convolution hyperparameter; when l = 1, is the truck travel feature matrix X obtained in step 1 s ;H s Tr H s The transposed matrix of s,v =[d(v)] is the degree matrix of node v at spatial scale s, which is used to describe the degree matrix of node v in the hypergraph G at the current spatial scale. s The internal element d(v) is the sum of the weights of all hyperedges connected to the vertex v, which is calculated by formula (6):

[0030]

[0031] D s,e =[d(e)] is the degree matrix of hyperedge e at spatial scale s, which is used to describe the degree matrix of hyperedge e in the hypergraph G at the current spatial scale. sThe internal element d(e) is the sum of the number of nodes connected by the hyperedge e, which is calculated by formula (7):

[0032]

[0033] According to the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by the present invention, the cross self-attention mechanism is used to integrate the implicit travel mode representations of the aforementioned trucks and their neighborhood groups at various spatial scales, and deep clustering is performed through a hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales, including:

[0034] According to step 3.1, the implicit travel mode representation of trucks in the lth convolutional layer The implicit travel mode representation of the truck neighborhood group in the lth convolutional layer in step 3.2 By introducing the cross self-attention mechanism, the implicit travel mode representation of the truck neighborhood group at the lth convolutional layer at the sth spatial scale is obtained through formula (8): Then, through formula (9), we can get a new expression of the travel mode of the truck neighborhood group under the lth convolution layer that integrates all spatial scales:

[0035]

[0036] In formula (8), CrossSelfAttention represents the cross self-attention mechanism; for The nth vector sequence of ; α mn is the mth query and the nth key The attention weight indicates the correlation between the two β mn is the mth query and the nth key The attention weight indicates the correlation between the two and is the attention scoring function, which is used to measure the correlation between the two in the formula and is quantitatively expressed using the dot product model;

[0037] Based on the fusion features of multiple spatial scales, the corresponding hypergraphs at each spatial scale are updated Corresponding splicing to obtain a dynamic hypergraph As the input of the hypergraph convolutional network at the convolutional layer l+1, and the output result corresponding to the autoencoder at the l+1 layer After cross-self-attention fusion, the hypergraph convolutional network operation is performed, and the probability distribution result of the truck travel category under the l+1th convolutional layer is obtained through the following formula (10): Finally, the dynamic classification of truck travel categories is completed;

[0038]

[0039] In formula (10), W′ s , H′ s , D′ s,v , D′ s,e Multiple spatial scale fusion hypergraphs The hyperedge weight set, incidence matrix, node degree matrix and hyperedge degree matrix of ;

[0040] For any spatial scale s, the implicit travel mode representation of trucks in the Lth convolutional layer is The distribution of the truck travel category is obtained by hypergraph clustering at the spatial scale s, and the cumulative error between the curve HC and the curve GT at all spatial scales is measured. Calculated by formula (11):

[0041]

[0042] Among them, HC sij It refers to the probability that truck i belongs to category j in curve HC under spatial scale s.

[0043] According to the method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering provided by the present invention, the training and testing of the truck travel category classification performance includes:

[0044] According to formula (12), based on the average reconstruction error and the and stated Get the overall error

[0045]

[0046] Among them, η1, η2 and η3 are the weight coefficients of the three errors respectively;

[0047] Repeat the above steps 3.1 to 3.3 repeatedly to minimize the cumulative error Until The value no longer changes, the model converges, and the truck travel categories obtained by classification are real and usable, completing the model training process;

[0048] The trained model is applied to the test dataset to verify the model’s performance in classifying truck travel categories at any spatial scale.

[0049] According to the truck travel classification method based on multi-scale hypergraph convolution deep clustering provided by the present invention, the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering is verified through comparison and ablation experiments, including:

[0050] The effective classification performance of the truck travel classification method based on multi-scale hypergraph convolutional deep clustering is verified through four indicators: accuracy, normalized virtual information, average rand index and F1-score.

[0051] The present invention also provides a truck travel classification system based on multi-scale hypergraph convolution deep clustering, comprising:

[0052] The truck travel feature matrix construction module is used to obtain the historical travel record set of N trucks in a given period based on the historical travel record set of N trucks; at the spatial scale s, F-dimensional travel features are extracted for each truck to construct the travel feature matrix X of N trucks. s ;

[0053] A truck travel mode hypergraph construction module is used to construct a truck travel mode hypergraph according to the truck travel feature matrix X s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel mode s ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s , H s It is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighborhood groups;

[0054] The implicit travel mode representation learning module for trucks is used to learn the implicit travel mode representation of the truck neighborhood group in the local space using the hypergraph convolutional network structure;

[0055] The neighborhood group implicit travel mode representation learning module is used to use the cross self-attention mechanism to integrate the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, and perform deep clustering through the hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales;

[0056] The hypergraph convolution deep clustering module is used to use the cross self-attention mechanism to integrate the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, and perform deep clustering through the hypergraph convolution network to obtain the classification results of truck travel categories at various spatial scales;

[0057] Training and testing modules, used to train and test the performance of truck travel classification;

[0058] The example verification module is used to apply the ACM dataset, Cite dataset and truck GPS dataset to verify the effective classification performance of the truck travel category classification method based on multi-scale hypergraph convolutional deep clustering through comparison and ablation experiments.

[0059] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements any of the above-mentioned methods for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering.

[0060] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering as described above is implemented.

[0061] The present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering.

[0062] In the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by the present invention, it is possible to rely on the fusion of hypergraph representation and hypergraph convolutional network technology to deeply model the implicit feature representation of the travel patterns of individual trucks and their neighborhood groups, and on the basis of considering the "many-to-many" association of multi-scale spatial interactions of travel patterns, to achieve accurate classification of truck groups, which is conducive to improving the level of refined management of managers for different categories of travel groups, so as to facilitate the subsequent implementation of classified policy formulation and refined emergency response plans. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0064] Figure 1 It is one of the flow diagrams of the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by an embodiment of the present invention;

[0065] Figure 2 This is a second flow chart of a method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided by an embodiment of the present invention;

[0066] Figure 3 It is a structural schematic diagram of a truck travel classification system based on multi-scale hypergraph convolution deep clustering provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. 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.

[0068] Figure 1 It is one of the flow charts of the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided in an embodiment of the present invention.

[0069] Figure 2 This is the second flow chart of the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided in an embodiment of the present invention.

[0070] like Figure 1 and Figure 2 As shown, this embodiment provides a method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering, including:

[0071] Step 1: Construct the travel feature matrix of trucks at each spatial scale. According to the historical travel record set of N trucks, obtain the historical travel record set of N trucks in a given period. At the spatial scale s, extract the F-dimensional travel features for each truck and construct the travel feature matrix of N trucks. The truck travel feature matrix X = {X1,…,X s ,…,X S}, where x s,i is the travel characteristic expression of the i-th truck at spatial scale s, 1≤i≤N; x i is the travel characteristic expression of the i-th truck, 1≤i≤N; the F-dimensional travel characteristics of each truck come from the time, space and attribute dimensions respectively. The time dimension includes the number of truck driver fatigue driving, night driving time, truck departure time and average daily travel time; the space dimension includes the truck travel distance, truck stop function and truck travel average speed; the attribute dimension includes license plate color and transportation trip number;

[0072] Step 2: construct a hypergraph of truck travel patterns at each spatial scale, model the many-to-many high-order association relationship of trucks, and calculate the truck travel feature matrix X according to step 1. s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel mode s ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s ;H s is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighboring groups, H s The internal element h(v,e) represents the affiliation relationship between node v and hyperedge e. The specific calculation method is shown in formula (1):

[0073]

[0074] Among them, v∈e means that the node is on the hyperedge;

[0075] Step 3: Build a multi-scale hypergraph convolutional deep clustering model to train and test the truck travel classification, including:

[0076] Step 3.1, using the autoencoder model, learn the implicit travel mode representation of trucks;

[0077] Step 3.2, using the hypergraph convolutional network structure, learn the implicit travel mode representation of the truck neighborhood group in the local space;

[0078] Step 3.3, using the cross self-attention mechanism, the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales are integrated, and deep clustering is performed through the hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales;

[0079] Step 3.4, training and testing the truck travel classification performance;

[0080] Step 4: Example verification, input the travel feature matrix X of N trucks at the S+1th spatial scale S+1 , construct a truck travel pattern hypergraph G′ that integrates multi-scale spatial features S+1 , realize the classification of truck travel groups; applied to the ACM dataset, Cite dataset and truck GPS dataset, through comparison and ablation experiments, verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering.

[0081] The trucks in this embodiment can be divided into different groups based on the above steps. During implementation, they can be divided based on the types of goods transported by the trucks. For example, they can be divided into trucks that transport vegetables, trucks that transport fruits, and trucks that transport garbage, etc. By classifying the trucks according to the goods they transport, it can help to carry out refined management of different truck categories and facilitate the formulation of classification policies.

[0082] In practical applications, in step 1, the information of the truck can be obtained from three dimensions: time, space, and attributes, and the category of the truck can be comprehensively judged from these three dimensions. The time information may include departure time, travel duration, etc. The departure time and travel duration may be different for different goods. For example, a truck transporting garbage may work from 5 to 6 in the morning or from 17 to 20 in the evening, while a truck transporting vegetables and fruits may need to work 24 hours a day to keep the goods fresh. Spatial information may include the truck's stop point and travel speed, etc. The stop point is generally the place where the truck loads and unloads. For different goods, The loading and unloading locations of trucks are definitely different. For example, a truck transporting vegetables and fruits may transport goods from a wholesale market to a vegetable market, or from a vegetable market to a supermarket, while a truck transporting garbage may transport goods from a residential area to a garbage station. In addition, the speed of the truck can also play an auxiliary role in judging the type of goods. For example, a truck transporting garbage may be slower for safety reasons, while a truck transporting vegetables and fruits may be faster for the freshness of the goods. In addition, for different trucks, their license plate colors also limit the types of goods that can be transported. The attribute information based on the license plate color can also assist in the classification of truck categories.

[0083] The hypergraph constructed in step 2 is a generalized graph structure. Compared with the traditional graph structure in which an edge can only connect two vertices, in the hypergraph structure, an edge can connect any number of vertices. The advantage of the hypergraph structure is that it can distinguish between the hierarchical and sub-hierarchical structures of information, and any complex information structure can be established by these two types of information.

[0084] A neighborhood is a topological structure on a set. A neighborhood is a special curve. If a point is the center point, any open interval is called the neighborhood of that point.

[0085] Clustering in step 3 is a process of grouping a collection of physical or abstract objects into classes of similar objects. A cluster is a collection of data objects that are similar to objects in the same cluster but different from objects in other clusters.

[0086] Specifically, in the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering provided in this embodiment, it can rely on the fusion of hypergraph representation and hypergraph convolutional network technology to deeply model the implicit feature representation of the travel patterns of individual trucks and their neighborhood groups. On the basis of considering the many-to-many correlation of multi-scale spatial interactions of travel patterns, it can achieve accurate classification of truck groups, which is conducive to improving the level of refined management of managers for different categories of travel groups, so as to facilitate the subsequent implementation of classified policy formulation and refined emergency response plans.

[0087] In an exemplary embodiment, the step of using an autoencoder model to learn an implicit travel mode representation of a truck includes:

[0088] The autoencoder model is introduced to learn the truck travel feature matrix X at the spatial scale s in step 1 through the built-in encoder. s Implicit travel mode representation at convolutional layer l (1≤l≤L) Calculated by formula (2):

[0089]

[0090] In the formula, σ ReLU is a linear activation function; AutoEncoder is an autoencoder. When l=1, The truck travel feature matrix X output in step 1 s ; W (l) is the convolution kernel weight of the lth layer, b (l) is the bias value of the lth layer, both of which are weight parameters of the model to be trained;

[0091] Through the decoder built into the autoencoder, after L convolutional layers, the reconstructed Complete the decoding operation; minimize the average reconstruction error produced by the autoencoder model at all spatial scales Formula (3) is used to express the truck travel mode that is closest to the actual situation:

[0092]

[0093] Among them, ||·|| Fro is the F norm;

[0094] For the spatial scale s, according to the truck travel characteristic matrix X s The distribution of truck travel categories in real scenarios is obtained by obtaining the probability distribution curve GT; according to the implicit feature representation of truck travel mode in the Lth convolutional layer The distribution of the truck travel category probability distribution curve TR obtained by the encoder division is obtained, and the cumulative error of the measurement curve GT and curve TR at all spatial scales is calculated. Calculated by formula (4):

[0095]

[0096] Among them, GT sij and AE sij are the probabilities that truck i belongs to category j in curve GT and curve AE at spatial scale s, respectively.

[0097] In an exemplary embodiment, the method of using a hypergraph convolutional network structure to learn the implicit travel mode representation of a truck neighborhood group in a local space includes:

[0098] According to the truck travel mode hypergraph G at any spatial scale s obtained in step 2 s , perform hypergraph convolutional network operations to capture the local spatial topological changes of the truck neighborhood group. For the lth convolutional layer, the implicit travel mode of the truck neighborhood group Calculated by formula (5):

[0099]

[0100] Where Θ is the convolution hyperparameter; when l = 1, is the truck travel feature matrix X obtained in step 1 s ;H s Tr H s The transposed matrix of s,v =[d(v)] is the degree matrix of node v at spatial scale s, which is used to describe the degree matrix of node v in the hypergraph G at the current spatial scale. s The internal element d(v) is the sum of the weights of all hyperedges connected to the vertex v, which is calculated by formula (6):

[0101]

[0102] D s,e =[d(e)] is the degree matrix of hyperedge e at spatial scale s, which is used to describe the degree matrix of hyperedge e in the hypergraph G at the current spatial scale. s The internal element d(e) is the sum of the number of nodes connected by the hyperedge e, which is calculated by formula (7):

[0103]

[0104] In the exemplary embodiment, the cross self-attention mechanism is used to fuse the implicit travel mode representations of the aforementioned trucks and their neighborhood groups at various spatial scales, and deep clustering is performed through a hypergraph convolutional network to obtain the division results of the truck travel categories at various spatial scales, including:

[0105] According to step 3.1, the implicit travel mode representation of trucks in the lth convolutional layer The implicit travel mode representation of the truck neighborhood group in the lth convolutional layer in step 3.2 By introducing the cross self-attention mechanism, the implicit travel mode representation of the truck neighborhood group at the lth convolutional layer at the sth spatial scale is obtained through formula (8): Then, through formula (9), we can get a new expression of the travel mode of the truck neighborhood group under the lth convolution layer that integrates all spatial scales:

[0106]

[0107]

[0108] In formula (8), CrossSelfAttention represents the cross self-attention mechanism; for The nth vector sequence of ; α mn is the mth query and the nth key The attention weight indicates the correlation between the two β mn is the mth query and the nth key The attention weight indicates the correlation between the two and is the attention scoring function, which is used to measure the correlation between the two in the formula and is quantitatively expressed using the dot product model;

[0109] Based on the fusion features of multiple spatial scales, the corresponding hypergraphs at each spatial scale are updated Corresponding splicing to obtain a dynamic hypergraph As the input of the hypergraph convolutional network at the convolutional layer l+1, and the output result corresponding to the autoencoder at the l+1 layer After cross-self-attention fusion, the hypergraph convolutional network operation is performed, and the probability distribution result of the truck travel category under the l+1th convolutional layer is obtained through the following formula (10): Finally, the dynamic classification of truck travel categories is completed;

[0110]

[0111] In formula (10), W′ s , H′ s , D′ s,v , D′ s,e Multiple spatial scale fusion hypergraphs The hyperedge weight set, incidence matrix, node degree matrix and hyperedge degree matrix of ;

[0112] For any spatial scale s, the implicit travel mode representation of trucks in the Lth convolutional layer is The distribution of the truck travel category is obtained by hypergraph clustering at the spatial scale s, and the cumulative error between the curve HC and the curve GT at all spatial scales is measured. Calculated by formula (11):

[0113]

[0114] Among them, HC sij It refers to the probability that truck i belongs to category j in curve HC under spatial scale s.

[0115] In an exemplary embodiment, the training and testing of truck travel classification performance includes:

[0116] According to formula (12), based on the average reconstruction error and the and stated Get the overall error

[0117]

[0118] Among them, η1, η2 and η3 are the weight coefficients of the three errors respectively;

[0119] Repeat the above steps 3.1 to 3.3 repeatedly to minimize the cumulative error Until The value no longer changes, the model converges, and the truck travel categories obtained by classification are real and usable, completing the model training process;

[0120] The trained model is applied to the test dataset to verify the model’s performance in classifying truck travel categories at any spatial scale.

[0121] In the exemplary embodiment, the comparison and ablation experiments are used to verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering, including:

[0122] The effective classification performance of the truck travel classification method based on multi-scale hypergraph convolutional deep clustering is verified through four indicators: accuracy, normalized virtual information, average rand index and F1-score.

[0123] The full name of accuracy is (Accuracy, ACC), the full name of normalized virtual information is (NormalizedMutual Information, NMI), the full name of average rand index is (Adjusted Rand Index, ARI), and the full name of F1-score is (F1-Score).

[0124] In practical applications, the larger the values ​​of the four indicators, the better the division performance.

[0125] In practical applications, the example verification method of the truck travel classification method based on multi-scale hypergraph convolution deep clustering includes the following steps:

[0126] (1) Working environment

[0127] The PyTorch framework is used to write the model code. All experimental codes are compiled and executed on Microsoft Windows 10 Home Chinese Edition and run on a hardware workstation equipped with an "11th Gen Intel(R) Core(TM) i5-11300H@3.10GHz" CPU and an "NVIDIA GeForce MX450(16GB RAM)" GPU graphics card.

[0128] This embodiment designs three spatial scales: site scale, block scale and administrative division scale.

[0129] (2) Introduction to experimental dataset

[0130] The experimental data sets of this embodiment are the ACM data set, the Cite data set, and the self-collected truck GPS data set. The ACM data set contains 1870 sample dimensions, 3025 samples, and 3 categories; the Cite data set contains 3703 sample dimensions, 3327 samples, and 6 categories; the self-collected truck GPS data set contains 6-dimensional travel feature data of 154,119 trucks after data preprocessing from June 1, 2019 to June 15, 2019.

[0131] (3) Benchmark model selection

[0132] In the comparison and ablation experiments, five baseline models are selected to compare the performance with the proposed model on a given dataset.

[0133] k-means clustering model: an unsupervised learning method that discovers hidden structures in data by dividing observations in a data set into different groups or clusters;

[0134] AE (Autoencoder): The built-in encoder maps the input data to a low-dimensional latent space, and then performs the clustering task in this space;

[0135] DBSCAN (Improved deep embedded clustering): a density-based clustering algorithm that can identify clusters of arbitrary shapes and is robust to outliers. This algorithm does not rely on a priori number of clusters and is suitable for exploring data sets with complex structures;

[0136] SDCN (Structured deep clustering network): constructs a k-nearest neighbor graph based on feature similarity, performs autoencoder pre-training, and captures the aggregation structure relationship between data;

[0137] AGCN (Attention-driven graph clustering network): extracts multi-scale node attributes and graph topology feature information, dynamically fuses them, and performs data clustering tasks;

[0138] (4) Experimental parameter setting

[0139] This embodiment uses the "AdamW" optimizer to perform 200 iterations of training on a given data set to carry out experiments on group segmentation performance research. In the experiments on the ACM data set, the learning rates of all models are set to 10-3; in the experiments on the Cite data set, the learning rates of all models are set to 10-4. In the experiments conducted on the ACM data set and the Cite data set, in order to ensure fair comparison of all methods, the present invention selects the recommended network configuration parameters according to each baseline model for optimization settings. The network size for AE and AGCN is set to "500-500-2000-10". In the experiments on the truck GPS data set, for k-means, we calculate the silhouette coefficient for different numbers of clusters, find the number of clusters with the highest silhouette coefficient of 3 as the optimal number of clusters, and output the silhouette coefficient value. DBSCAN finds the optimal parameter combination of eps=1.1 and MinPts=7 through the strategy of number tuning. The network size of the AE autoencoder is set to "500-500-2000-10". The network sizes of the built-in encoder of SDCN and the GCN module are set to "500-500-2000-10". In the method proposed in the present invention, the weight coefficients η1, η2 and η3 in formula (12) are set to 1, 0.1 and 0.01 respectively.

[0140] (5) Selection of evaluation indicators

[0141] The experiments conducted on the ACM dataset and the Cite dataset evaluate the clustering performance of each model through four common indicators, namely, Accuracy (ACC), Normalized Virtual Information (NMI), Average Rand Index (ARI), and F1-score (F1). Generally speaking, the larger the evaluation index, the better the effect.

[0142] In the experiment conducted on the truck GPS dataset, the silhouette coefficient method was used to evaluate the clustering performance of each model. The larger the evaluation index, the better the effect.

[0143] (6) Experimental results and analysis

[0144] Table 1 shows the experimental results of group partitioning comparison of various models on two public datasets, ACM and Cite. In general, the proposed method achieves the best clustering performance (see the bold data in Table 1). Compared with the baseline model SDCN with the second best group partitioning performance, the proposed method has achieved rapid growth in various indicators on the two public datasets, namely: ACC increased by 1.62% and 2.47%, NMI increased by 4.99% and 1.61%, ARI increased by 3.86% and 2.86%, and F1 increased by 1.63% and 0.36%, respectively, which proves that the present invention has good group partitioning performance. This is attributed to the fact that the proposed method of the present invention has a hypergraph module and a feature fusion module. The former can fully learn the similarity of implicit travel patterns between individuals and domain groups; the latter can effectively combine the travel pattern characteristics of the individual itself and its neighborhood groups, better describe the implicit clustering characteristics of the target individual, and thus improve the performance of group partitioning.

[0145] Table 1 Comparative experimental results of various models on public datasets

[0146]

[0147] Table 2 reflects the improvement of the group division performance by the hypergraph module and feature fusion module of the proposed method. The second row of experimental results for each data set in the table is the distribution of group division performance when the proposed method only carries the first module; the third row of experimental results for each data set in the table is the distribution of group division performance when the proposed method carries the above two modules. As can be seen from the table, adding the hypergraph module can increase the ACC of the proposed model in the two data sets by 0.4% and 2.72%, NMI by 0.23% and 2.9%, ARI by 3.22% and 3.15%, and F1 index by a decrease, indicating that only constructing the hypergraph module may not be sensitive on the public data set, while other indexes have increased, confirming that the hypergraph module can effectively explain the complex many-to-many association characteristics between individuals; after adding the feature fusion module, the proposed model can increase the ACC by 0.43% and 0.8%, NMI by 5.07% and 2.35%, ARI by 6.08% and 3.07%, and F1 by 2.61% and 2.52% on the basis of the hypergraph module. This fully shows that the design of the hypergraph module and the feature fusion module of the present invention are both helpful to improve the performance of accurate group division.

[0148] Table 2 Ablation test results of the built-in modules of the proposed method on public datasets

[0149]

[0150] Table 3 shows the group segmentation effect of the proposed method in the truck field. Compared with the traditional algorithms K-means and DBSCAN, the group segmentation effect of the proposed method is improved by about 8.91% and 4.68% respectively; compared with the algorithms AE, SDCN and AGCN that introduce deep learning technology, the group segmentation effect is improved by about 8.14%, about 4.00% and about 3.02% respectively. It can be seen that the proposed method can not only effectively process the high-order relations in the truck travel data, but also provide more accurate group segmentation effect than traditional clustering algorithms and other deep learning methods, thereby more accurately classifying truck travel categories.

[0151] Table 3 Effect of the proposed method on truck travel mode classification on self-collected truck GPS dataset

[0152]

[0153] The following is a description of the truck travel category classification system based on multi-scale hypergraph convolution deep clustering provided by the present invention. The truck travel category classification system based on multi-scale hypergraph convolution deep clustering described below and the truck travel category classification method based on multi-scale hypergraph convolution deep clustering described above can be referenced to each other.

[0154] Figure 3 It is a structural schematic diagram of a truck travel classification system based on multi-scale hypergraph convolution deep clustering provided by an embodiment of the present invention.

[0155] like Figure 3 As shown, the truck travel classification system based on multi-scale hypergraph convolution deep clustering provided in this embodiment includes:

[0156] The truck travel feature matrix construction module 301 is used to obtain the historical travel record set of N trucks in a given period according to the historical travel record set of N trucks; extract F-dimensional travel features for each truck at the spatial scale s, and construct the travel feature matrix X of N trucks. s ;

[0157] Truck travel mode hypergraph construction module 302 is used to construct a hypergraph based on the truck travel feature matrix X s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel modes ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s , H s It is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighborhood groups;

[0158] The implicit travel mode representation learning module 303 of trucks is used to learn the implicit travel mode representation of the truck neighborhood group in the local space by using the hypergraph convolutional network structure;

[0159] The neighborhood group implicit travel mode representation learning module 304 is used to use the cross self-attention mechanism to integrate the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, perform deep clustering through a hypergraph convolutional network, and obtain the classification results of truck travel categories at various spatial scales;

[0160] The hypergraph convolution deep clustering module 305 is used to utilize the cross self-attention mechanism to fuse the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, perform deep clustering through the hypergraph convolution network, and obtain the classification results of truck travel categories at various spatial scales;

[0161] A training and testing module 306, for training and testing the performance of truck travel classification;

[0162] The example verification module 307 is used to apply the ACM dataset, the Cite dataset and the truck GPS dataset to verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering through comparison and ablation experiments.

[0163] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. 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 truck travel classification method based on multi-scale hypergraph convolutional deep clustering, characterized in that: include: Step 1: Construct the travel feature matrix of trucks at each spatial scale. According to the historical travel record set of N trucks, obtain the historical travel record set of N trucks in a given period. At the spatial scale s, extract the F-dimensional travel features for each truck and construct the travel feature matrix of N trucks. The truck travel feature matrix X = {X1,…,X s ,…,X S }, where x s,i is the travel characteristic expression of the i-th truck at spatial scale s, 1≤i≤N; x i is the travel characteristic expression of the i-th truck, 1≤i≤N; the F-dimensional travel characteristics of each truck come from the time, space and attribute dimensions respectively. The time dimension includes the number of truck driver fatigue driving, night driving time, truck departure time and average daily travel time; the space dimension includes the truck travel distance, truck stop function and truck travel average speed; the attribute dimension includes license plate color and transportation trip number; Step 2: construct a hypergraph of truck travel patterns at each spatial scale, model the many-to-many high-order association relationship of trucks, and calculate the truck travel feature matrix X according to step 1. s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel mode s ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s ;H s is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighboring groups, H s The internal element h(v,e) represents the affiliation relationship between node v and hyperedge e. The specific calculation method is shown in formula (1): Among them, v∈e means that the node is on the hyperedge; Step 3: Build a multi-scale hypergraph convolutional deep clustering model to train and test the truck travel classification, including: Step 3.1, using the autoencoder model, learn the implicit travel mode representation of trucks; Step 3.2, using the hypergraph convolutional network structure, learn the implicit travel mode representation of the truck neighborhood group in the local space; Step 3.3, using the cross self-attention mechanism, the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales are integrated, and deep clustering is performed through the hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales; Step 3.4, training and testing the truck travel classification performance; Step 4: Example verification, input the travel feature matrix X of N trucks at the S+1th spatial scale S+1 , construct a truck travel pattern hypergraph G′ that integrates multi-scale spatial features S+1 , realize the classification of truck travel groups; applied to the ACM dataset, Cite dataset and truck GPS dataset, through comparison and ablation experiments, verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering.

2. The method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering according to claim 1 is characterized in that: The method of using the autoencoder model to learn the implicit travel mode representation of trucks includes: The autoencoder model is introduced to learn the truck travel feature matrix X at the spatial scale s in step 1 through the built-in encoder. s Implicit travel mode representation at convolutional layer l (1≤l≤L) Calculated by formula (2): In the formula, σ ReLU is a linear activation function; AutoEncoder is an autoencoder. When l=1, The truck travel feature matrix X output in step 1 s ; W (l) is the convolution kernel weight of the lth layer, b (l) is the bias value of the lth layer, both of which are weight parameters of the model to be trained; Through the decoder built into the autoencoder, after L convolutional layers, the reconstructed Complete the decoding operation; minimize the average reconstruction error produced by the autoencoder model at all spatial scales Formula (3) is used to express the truck travel mode that is closest to the actual situation: Among them, ||·|| Fro is the F norm; For the spatial scale s, according to the truck travel characteristic matrix X s The distribution of truck travel categories in real scenarios is obtained by obtaining the probability distribution curve GT; according to the implicit feature representation of truck travel mode in the Lth convolutional layer The distribution of the truck travel category probability distribution curve TR obtained by the encoder division is obtained, and the cumulative error of the measurement curve GT and curve TR at all spatial scales is calculated. Calculated by formula (4): Among them, GT sij and AE sij are the probabilities that truck i belongs to category j in curve GT and curve AE at spatial scale s, respectively.

3. The method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering according to claim 1 is characterized in that: The hypergraph convolutional network structure is used to learn the implicit travel mode representation of the truck neighborhood group in the local space, including: According to the truck travel mode hypergraph G at any spatial scale s obtained in step 2 s , perform hypergraph convolutional network operations to capture the local spatial topological changes of the truck neighborhood group. For the lth convolutional layer, the implicit travel mode of the truck neighborhood group Calculated by formula (5): Where Θ is the convolution hyperparameter; when l = 1, is the truck travel feature matrix X obtained in step 1 s ;H s Tr H s The transposed matrix of s,v =[d(v)] is the degree matrix of node v at spatial scale s, which is used to describe the degree matrix of node v in the hypergraph G at the current spatial scale. s The internal element d(v) is the sum of the weights of all hyperedges connected to the vertex v, which is calculated by formula (6): D s,e =[d(e)] is the degree matrix of hyperedge e at spatial scale s, which is used to describe the degree matrix of hyperedge e in the hypergraph G at the current spatial scale. s The internal element d(e) is the sum of the number of nodes connected by the hyperedge e, which is calculated by formula (7):

4. The method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering according to claim 2 is characterized in that: The cross self-attention mechanism is used to integrate the implicit travel mode representations of the aforementioned trucks and their neighborhood groups at various spatial scales, and deep clustering is performed through a hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales, including: According to step 3.1, the implicit travel mode representation of trucks in the lth convolutional layer The implicit travel mode representation of the truck neighborhood group in the lth convolutional layer in step 3.2 By introducing the cross self-attention mechanism, the implicit travel mode representation of the truck neighborhood group at the lth convolutional layer at the sth spatial scale is obtained through formula (8): Then, through formula (9), we can get a new expression of the travel mode of the truck neighborhood group under the lth convolution layer that integrates all spatial scales: In formula (8), CrossSelfAttention represents the cross self-attention mechanism; for The nth vector sequence of ; α mn is the mth query and the nth key The attention weight indicates the correlation between the two β mn is the mth query and the nth key The attention weight indicates the correlation between the two and is the attention scoring function, which is used to measure the correlation between the two in the formula and is quantitatively expressed using the dot product model; Based on the fusion features of multiple spatial scales, the corresponding hypergraphs at each spatial scale are updated Corresponding splicing to obtain a dynamic hypergraph As the input of the hypergraph convolutional network at the convolutional layer l+1, and the output result corresponding to the autoencoder at the l+1 layer After cross-self-attention fusion, the hypergraph convolutional network operation is performed, and the probability distribution result of the truck travel category under the l+1th convolutional layer is obtained through the following formula (10): Finally, the dynamic classification of truck travel categories is completed; In formula (10), W′ s , H′ s , D′ s,v , D′ s,e Multiple spatial scale fusion hypergraphs The hyperedge weight set, incidence matrix, node degree matrix and hyperedge degree matrix of ; For any spatial scale s, the implicit travel mode representation of trucks in the Lth convolutional layer is The distribution of the truck travel category is obtained by hypergraph clustering at the spatial scale s, and the cumulative error between the curve HC and the curve GT at all spatial scales is measured. Calculated by formula (11): Among them, HC sij It refers to the probability that truck i belongs to category j in curve HC under spatial scale s.

5. The method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering according to claim 4 is characterized in that: The training and testing of the truck travel classification performance includes: According to formula (12), based on the average reconstruction error and the and stated Get the overall error Among them, η1, η2 and η3 are the weight coefficients of the three errors respectively; Repeat the above steps 3.1 to 3.3 to minimize the cumulative error Until The value no longer changes, the model converges, and the truck travel categories obtained by classification are real and usable, completing the model training process; The trained model is applied to the test dataset to verify the model’s performance in classifying truck travel categories at any spatial scale.

6. The method for classifying truck travel categories based on multi-scale hypergraph convolutional deep clustering according to claim 1 is characterized in that: The comparison and ablation experiments verify the effective classification performance of the truck travel classification method based on multi-scale hypergraph convolution deep clustering, including: The effective classification performance of the truck travel classification method based on multi-scale hypergraph convolutional deep clustering is verified through four indicators: accuracy, normalized virtual information, average rand index and F1-score.

7. A truck travel classification system based on multi-scale hypergraph convolutional deep clustering, characterized by: include: The truck travel feature matrix construction module is used to obtain the historical travel record set of N trucks in a given period based on the historical travel record set of N trucks; at the spatial scale s, F-dimensional travel features are extracted for each truck to construct the travel feature matrix X of N trucks. s ; A truck travel mode hypergraph construction module is used to construct a truck travel mode hypergraph according to the truck travel feature matrix X s , construct X s The travel pattern hypergraph G at spatial scale s s =(V s ,X s ,W s ,E s ,H s ), take N trucks as N nodes on the graph, and form a node set V that represents all trucks s ; The F-dimensional travel characteristics of each truck are regarded as the attributes of each node, forming a node attribute set X that expresses the truck travel mode s ; Connect the two nodes of the nearest neighbor, and use the inverse of the Euclidean distance between them as the edge weight of the corresponding node, forming an edge weight set W that expresses the importance of the neighborhood group s ; Select the first k-1 neighboring trucks with the smallest Euclidean distance to form the neighborhood group of the i-th truck By analogizing the above neighborhood groups to hyperedges, we obtain M hyperedge sets, which constitute the hyperedge set E that expresses the complex association relationship of many-to-many travel modes of neighborhood groups. s , H s It is an N-row and N-column association matrix expressing the affiliation relationship between a single truck and its neighborhood groups; The implicit travel mode representation learning module for trucks is used to learn the implicit travel mode representation of the truck neighborhood group in the local space using the hypergraph convolutional network structure; The neighborhood group implicit travel mode representation learning module is used to use the cross self-attention mechanism to integrate the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, and perform deep clustering through the hypergraph convolutional network to obtain the classification results of truck travel categories at various spatial scales; The hypergraph convolution deep clustering module is used to use the cross self-attention mechanism to integrate the implicit travel mode representations of trucks and their neighborhood groups at various spatial scales, and perform deep clustering through the hypergraph convolution network to obtain the classification results of truck travel categories at various spatial scales; Training and testing modules, used to train and test the performance of truck travel classification; The example verification module is used to apply the ACM dataset, Cite dataset and truck GPS dataset to verify the effective classification performance of the truck travel category classification method based on multi-scale hypergraph convolutional deep clustering through comparison and ablation experiments.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering as described in any one of claims 1-6 is implemented.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering as described in any one of claims 1 to 6 is implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for classifying truck travel categories based on multi-scale hypergraph convolution deep clustering as described in any one of claims 1 to 6 is implemented.

Images