Freight traffic prediction method and device based on topological network spatio-temporal evolution

Abstract

The invention discloses a freight flow prediction method and device based on topological network spatio-temporal evolution, and the method comprises the steps: constructing a topological network based on the road information of a to-be-predicted traffic road section, and determining an original flow feature vector corresponding to each road section in the topological network according to the historical flow data corresponding to the to-be-predicted traffic road section; performing influence intensity feature correction on the original traffic feature vector based on the influence relationship among the road sections in the topological network to obtain a corrected traffic feature vector; performing feature prediction on the corrected flow feature vector by using a time sequence model to obtain a predicted traffic state characterization vector; and determining the predicted traffic flow information corresponding to the to-be-predicted traffic road section based on the predicted traffic state representation vector, so that the time sequence relationship and the topological structure relationship of the freight flow can be considered at the same time, the change trend of the freight flow in time and space can be effectively captured, and the accuracy and reliability of freight flow prediction can be improved.

Description

Technical Field

[0001] The present disclosure generally relates to the field of traffic forecasting, and in particular to a method and device for freight flow forecasting based on spatiotemporal evolution of a topological network. Background Art

[0002] To achieve real-time monitoring and management of regional freight channels, it is necessary not only to integrate the hardware of the monitoring and detection systems of the entire region, but also to realize the unified dispatch of regional freight channel traffic information. The realization of the freight channel traffic information dispatch function requires the support of the dynamic forecast data of the traffic flow of the regional freight channel, so as to realize the detection of freight flow information, emergency event detection, and the prediction and analysis of road network operation status. Therefore, one of the key issues in realizing the traffic dispatch command function of the freight channel is the accurate regional freight channel traffic forecast.

[0003] At present, there are relatively few studies on the traffic flow of regional transport corridors dominated by freight traffic. The traditional freight traffic forecasting model is limited by the complexity of research data and freight activities, resulting in low accuracy. Domestic freight forecasting models often start from a more macro perspective, using economic and social factors to infer regional freight demand, and freight corridor planning usually relies on traditional survey data with small data volume, complex acquisition channels and low accuracy, and lacks a grasp of freight characteristics. Summary of the invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, it is desired to provide a freight flow prediction method and device based on the spatiotemporal evolution of topological networks, which can simultaneously take into account the time series relationship and topological structure relationship of freight flow, effectively capture the changing trends of freight flow in time and space, and improve the accuracy and reliability of freight flow prediction.

[0005] In a first aspect, an embodiment of the present application provides a method for predicting freight flow based on spatiotemporal evolution of a topological network, comprising:

[0006] A topological network is constructed based on the road information of the traffic section to be predicted, and an original traffic feature vector corresponding to each section in the topological network is determined according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix; the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which the section belongs;

[0007] Based on the influence relationship between the road sections in the topological network, the original flow characteristic vector is corrected by the influence intensity characteristic to obtain a corrected flow characteristic vector;

[0008] Using a time series model to perform feature prediction on the corrected traffic feature vector to obtain a predicted traffic state representation vector;

[0009] Based on the predicted traffic state characterization vector, the predicted traffic flow information corresponding to the traffic section to be predicted is determined.

[0010] In some embodiments, determining the original traffic feature vector corresponding to each road section in the topological network according to the historical traffic data corresponding to the traffic road section to be predicted includes:

[0011] Constructing a traffic matrix based on the historical traffic data corresponding to the traffic section to be predicted;

[0012] Determining a periodic flow curve corresponding to the historical flow data according to the flow matrix;

[0013] Based on the periodic flow curve and the historical flow data, construct the three-dimensional traffic flow information matrix;

[0014] Based on the three-dimensional traffic flow information matrix, the original traffic flow feature vector corresponding to each road section in the topological network is determined.

[0015] In some embodiments, it also includes:

[0016] A three-dimensional traffic flow information matrix is ​​constructed based on the periodic flow curve and the disturbance residual of the historical flow data.

[0017] In some embodiments, the modifying the original flow feature vector based on the influence relationship between the road sections in the topological network to obtain the modified flow feature vector includes:

[0018] Based on the self-matching-shared matching mechanism, calculating the influence intensity coefficient between the first-order adjacent sections in the topological network;

[0019] Based on the impact intensity coefficient, the original flow characteristic vector is corrected by impact intensity characteristics to obtain a corrected flow characteristic vector.

[0020] In some embodiments, the modifying the original flow characteristic vector based on the influence intensity coefficient to obtain the modified flow characteristic vector includes:

[0021] Normalizing the impact intensity coefficient;

[0022] Using the normalized impact intensity coefficients of K dimensions, respectively correcting the impact intensity characteristics of the original traffic feature vector;

[0023] The corrected average value of the K dimensions is used as the corrected traffic feature vector.

[0024] In some embodiments, the method of using a time series model to perform feature prediction on the corrected traffic feature vector to obtain a predicted traffic state representation vector includes:

[0025] The modified traffic feature vector is input into the time series model for feature prediction to obtain a predicted traffic state representation vector.

[0026] In some embodiments, the step of determining the predicted traffic flow information corresponding to the traffic section to be predicted based on the predicted traffic state representation vector includes:

[0027] Decoding the predicted traffic state representation vector based on the regression relationship between the predicted characteristic parameter and the predicted value traffic state to obtain a predicted output value corresponding to the traffic section to be predicted;

[0028] Based on the periodic flow curve and the predicted output value, the predicted traffic flow information corresponding to the traffic section to be predicted is determined.

[0029] In a second aspect, an embodiment of the present application provides a freight flow prediction device based on the spatiotemporal evolution of a topological network, comprising:

[0030] A first determination module is used to construct a topological network based on the road information of the traffic section to be predicted, and determine the original traffic feature vector corresponding to each section in the topological network according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix; the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which the section belongs;

[0031] A correction module, used for correcting the influence intensity characteristics of the original flow characteristic vector based on the influence relationship between the road sections in the topological network to obtain a corrected flow characteristic vector;

[0032] A prediction module, used for performing feature prediction on the corrected traffic feature vector using a time series model to obtain a predicted traffic state representation vector;

[0033] The second determination module is used to determine the predicted traffic flow corresponding to the traffic section to be predicted based on the predicted traffic state representation vector.

[0034] In a third aspect, an embodiment of the present application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method described in the embodiment of the present application when executing the program.

[0035] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the embodiment of the present application.

[0036] The freight flow prediction method based on the spatiotemporal evolution of the topological network proposed in the embodiment of the present application constructs the original flow feature vector using a three-dimensional traffic flow information matrix that expresses more traffic flow differences, converts the traffic flow into flow difference information with a finer time granularity, realizes the prediction of the intensity of flow difference changes, improves the attention to differential flow, and avoids the edge loss caused by the overall flow prediction, thereby improving the accuracy of finer-grained flow prediction.

[0037] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0039] Figure 1 A schematic diagram of a flow chart of a freight flow prediction method based on spatiotemporal evolution of a topological network provided in an embodiment of the present application is shown;

[0040] Figure 2 A schematic diagram of a flow chart of a method for predicting freight flow based on spatiotemporal evolution of a topological network provided by another embodiment of the present application is shown;

[0041] Figure 3 A schematic diagram of a flow chart of a freight flow prediction method based on spatiotemporal evolution of a topological network provided by another embodiment of the present application is shown;

[0042] Figure 4 A schematic diagram of a flow chart of a freight flow prediction method based on spatiotemporal evolution of a topological network provided in yet another embodiment of the present application is shown;

[0043] Figure 5 A schematic diagram of the structure of a freight flow prediction device based on spatiotemporal evolution of a topological network provided in one embodiment of the present application is shown;

[0044] Figure 6 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown. DETAILED DESCRIPTION

[0045] The present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the relevant invention, rather than to limit the invention. It is also necessary to explain that, for ease of description, only the parts related to the invention are shown in the accompanying drawings.

[0046] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0047] In order to further illustrate the technical solution provided by the embodiment of the present application, this is described in detail below in conjunction with the accompanying drawings and specific implementation methods. Although the embodiment of the present application provides the method operation instruction steps shown in the following embodiments or drawings, more or less operation instruction steps may be included in the method based on routine or no creative labor. In the steps where there is no necessary causal relationship logically, the execution order of these steps is not limited to the execution order provided by the embodiment of the present application. The method may be executed in the order of the method shown in the embodiment or drawings or in parallel during the actual processing process or when the device is executed.

[0048] Please refer to Figure 1 , Figure 1 FIG. 1 is a flow chart of a method for predicting freight flow based on the spatiotemporal evolution of a topological network provided by an embodiment of the present application. Figure 1 As shown, the method includes:

[0049] Step 100, constructing a topological network based on the road information of the traffic section to be predicted, and determining the original traffic feature vector corresponding to each section in the topological network according to the historical traffic data corresponding to the traffic section to be predicted.

[0050] That is to say, in the embodiment of the present application, the road network to which the traffic section to be predicted belongs is constructed as a topological network. Specifically, the road section is abstracted as a node in the topological network, and the topological network G = (V, B) corresponding to the road network is obtained, where V is a vertex set, V = {v1, v2, ..., v N},v i represents the i-th road section, N represents the number of roads, and B represents the adjacency matrix in the road network, which represents the connection relationship between each road section. It can also be expressed as follows:

[0051]

[0052] In a feasible embodiment, according to the historical traffic data corresponding to the traffic section to be predicted, determining the original traffic feature vector corresponding to each section in the topological network includes:

[0053] Step 110, constructing a traffic matrix based on the historical traffic data corresponding to the traffic section to be predicted.

[0054] It should be noted that the flow matrix is ​​the flow matrix of each road section in the historical number of days. For example, for N road sections in the topological network, the flow matrix S corresponding to the truck flow data in the historical 7 days is n for:

[0055]

[0056] in, represents the freight flow of N sections on the nth day, Represents the freight flow of N sections at time t on the nth day, where t is the length of the daily time series.

[0057] Step 120: Determine the periodic flow curve corresponding to the historical flow data according to the flow matrix.

[0058] It should be noted that the periodic flow curve is the flow curve of the traffic section to be predicted within the historical time period (7 days in the embodiment of the present application). It should also be noted that in the subsequent flow prediction of the present application, the freight flow is also predicted on the basis of the periodic flow curve. In other words, in the embodiment of the present application, it is believed that freight traffic has a certain periodic change law. The embodiment of the present application divides the historical flow data according to the periodicity to establish a corresponding flow matrix, and then determines the periodic flow curve, which can better reflect the flow change of freight flow within a time period.

[0059] For example, according to the traffic matrix S n , for each time unit, construct the periodic flow curve corresponding to each road section. Specifically, the average flow rate at each time point in a cycle (within 7 days) can be calculated as the periodic flow curve of the cycle:

[0060]

[0061] D i =[D i (1) D i (2) … D i (t)]

[0062] Among them, D i (t) is the average flow value of section i at time t, i = 1, 2, ..., N, is the freight flow of section i at time t on the nth day, D i is the periodic flow curve of section i.

[0063] Step 130, constructing a three-dimensional traffic flow information matrix based on the periodic traffic curve and historical traffic data.

[0064] Preferably, a three-dimensional traffic flow information matrix is ​​constructed based on the periodic flow curve and the disturbance residual of the historical flow data.

[0065] It should be noted that, in the embodiment of the present application, the disturbance residual is the disturbance magnitude of the actual historical flow data of each section at each time compared with the periodic flow curve trend, which can be specifically expressed as:

[0066]

[0067] in, is a three-dimensional traffic flow information matrix, which represents the disturbance residual of the road section i at time t on the nth day in the cycle relative to the periodic flow curve.

[0068] Step 140, based on the three-dimensional traffic flow information matrix, determine the original traffic flow feature vector corresponding to each road section in the topological network.

[0069] It should be noted that, in the embodiment of the present application, the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix, wherein the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which a road section belongs.

[0070] Among them, the time intervals of the first time dimension, the second time dimension and the original traffic feature vector can be the same or different. It should be understood that the construction of the three-dimensional traffic flow information matrix can fully express the periodic characteristics of the first time dimension itself and the periodic characteristics of the second time dimension by utilizing the independence of the two time dimensions, as well as the dependence of the second time dimension on the first time dimension, so as to improve the expression ability of the feature vector in the time domain when constructing the original traffic feature vector, and provide richer and more reliable time domain information for the subsequent use of the time series model.

[0071] Specifically, based on the three-dimensional traffic flow information matrix Determine the time interval length T used for prediction and obtain the original traffic feature vector

[0072]

[0073] Where T is the length of the time interval, is the original traffic feature vector of road section i, F is the number of features for each node.

[0074] It should be noted that the time interval length T is a time interval limited to a historical cycle, that is, the time interval length is the time interval for segmenting the second time dimension. Therefore, the original traffic feature vector realizes time series segmentation but still retains the periodic expression of the first time dimension. Moreover, by segmenting the second time dimension according to the demand, the time accuracy can be set according to the prediction demand, such as hours, which can effectively improve the accuracy of the prediction in the time dimension while ensuring the overall prediction reliability.

[0075] It should be understood that for each road segment in the topological network, there are:

[0076]

[0077] Among them, Y is the set of original traffic feature vectors corresponding to the topological network, N is the number of road sections in the topological network, is the original traffic feature vector of road section i, i=1,2,…,N.

[0078] Therefore, the embodiment of the present application utilizes the topological network and historical traffic data of the traffic section to be predicted to determine the three-dimensional traffic flow information matrix corresponding to each section, and then determines the original traffic flow feature vector used for traffic flow prediction, so that the original traffic flow feature vector used for traffic flow prediction fully includes the historical information of freight traffic flow information in the time dimension and space dimension, effectively avoiding the difference problem caused by the time prediction and space prediction of freight traffic flow respectively, thereby improving the accuracy and robustness of the prediction results of the post-prediction model.

[0079] Moreover, the present application utilizes the traffic matrix, three-dimensional traffic flow information matrix and original traffic feature vector to effectively construct the hourly data volume of the freight traffic network, further improving the spatiotemporal accuracy of freight traffic prediction.

[0080] Step 200, based on the influence relationship between each road section in the topological network, the original flow characteristic vector is corrected by the influence intensity characteristic to obtain a corrected flow characteristic vector.

[0081] It should be noted that due to the upstream and downstream influence relationships of freight traffic flows, such as converging sections, diverging sections, etc., this application uses the topological network node importance theory to correct and strengthen the original flow feature vector in the spatial dimension, so as to better pay attention to the impact of adjacent sections on the flow prediction of the predicted traffic sections, and improve the accuracy and reliability of the final predicted traffic status.

[0082] In a possible embodiment, if Figure 3As shown, step 200, based on the influence relationship between each road section in the topological network, the original flow feature vector is corrected by the influence intensity feature to obtain the corrected flow feature vector, including:

[0083] Step 210 , based on the self-matching-shared matching mechanism, calculate the influence intensity coefficient between the first-order adjacent sections in the topological network.

[0084] It should be understood that, due to the connection relationship of traffic sections, adjacent sections with a first-order adjacency relationship have a stronger flow impact, such as extending from an upstream section to a downstream section, or an upstream section diverting to a downstream section, or multiple upstream sections converging to a downstream section, etc. Therefore, the embodiments of the present application focus on the influence intensity coefficient between first-order adjacent sections in a topological network.

[0085] Among them, the influence intensity coefficient can be expressed as:

[0086]

[0087] Among them, e ij is the influence intensity coefficient, representing the influence of road section i on road section j, a:R F′ ×R F →R is a self-matching-shared matching mechanism, which is used to implement a single-layer feedforward neural network, W∈R F′×F is the linear transformation parameter matrix of the science department.

[0088] Step 220, based on the impact intensity coefficient, the original flow characteristic vector is corrected according to the impact intensity characteristic to obtain a corrected flow characteristic vector.

[0089] In a specific embodiment, step 220, based on the impact intensity coefficient, the original flow feature vector is subjected to impact intensity feature correction to obtain a corrected flow feature vector, including:

[0090] Step 221, normalizing the influence intensity coefficient.

[0091] Optionally, in an embodiment of the present application, a softmax function is used to normalize all choices of j, and the weights are activated by a LeakyReLU function:

[0092]

[0093] Among them, α ij is the normalized influence intensity coefficient, . T represents transposition, || represents the connection operation, N is the number of road segments, N i is the neighbor node of road segment i, is the parameterized matrix,

[0094] Step 222, using the normalized impact intensity coefficient, respectively correct the impact intensity characteristics of the original flow feature vectors.

[0095] Specifically, the influence intensity coefficient is used to correct the influence intensity characteristics of the original traffic feature vector, so that the information of the first-order adjacent sections can be aggregated in the original traffic feature vector of each section. The importance of information of different first-order neighbor sections can also be dynamically determined according to the influence intensity coefficient, so that more attention can be paid to the adjacent sections that are helpful to the current task in post-prediction.

[0096] Exemplarily, the original traffic feature vector can be modified using the following expression:

[0097]

[0098] Among them, α ij is the normalized impact intensity coefficient, is the modified traffic feature vector of section i after modification in this dimension, is the original traffic feature vector of the first-order adjacent road segment, and σ(·) is the sigmoid activation function.

[0099] Step 223: The corrected average value of the K dimensions is used as the corrected traffic feature vector.

[0100] It should be noted that the K dimensions are attention coefficients set when modifying the influence strength based on the topological network graph.

[0101] In the related art, K ​​attention layers are usually superimposed to enhance the attention of the traffic section to be predicted to the first-order adjacent sections, that is, the K correction results obtained by analyzing multiple attention layers are spliced ​​to achieve full attention to the first-order adjacent sections. However, this processing method will cause feature redundancy in the corrected traffic feature vector and increase the data processing pressure of the subsequent prediction model. Moreover, the three-dimensional traffic flow information used in this application to generate the original traffic feature vector has a better differential information expression and contains less general information. Based on this, this application further proposes to use the corrected average value of K dimensions as the corrected traffic feature vector.

[0102] That is to say, the present application constructs a three-dimensional traffic flow information matrix using the disturbance residual of the periodic traffic curve and the historical traffic data, and based on the original traffic feature vector obtained therefrom, the original traffic feature vector contains more differential information of the traffic flow of the traffic section to be predicted. Based on this, when the original traffic feature vector is corrected for the impact intensity based on the first-order adjacent sections of the traffic section to be predicted, more attention can be paid to the impact intensity, that is, there is no need to splice all the attention analysis results, and the impact intensity correction requirements can be met only by the average value.

[0103] Exemplarily, the following formula can be used to implement it:

[0104]

[0105] Among them, K is the attention coefficient set when modifying the influence strength based on the topological network graph, α ij is the normalized impact intensity coefficient, is the modified traffic characteristic vector corresponding to section i, is the original traffic feature vector of the first-order adjacent road segment, σ(·) is the sigmoid activation function, W k is the linear transformation parameter matrix corresponding to the kth dimension.

[0106] Therefore, the freight flow prediction method based on the spatiotemporal evolution of the topological network proposed in the embodiment of the present application adds the influence intensity correction between the sections based on the topological network to the original flow feature vector before the flow prediction, so that the corrected flow feature vector can effectively reflect the correlation between the sections in the spatial domain, realize the expression of the flow influence between the upstream and downstream sections on the feature vector, and improve the mapping of the corrected flow characteristics to the spatial information. At the same time, the corrected average value is used as the final corrected flow feature vector to achieve the purpose of effectively reducing redundant data while ensuring the correction of the influence intensity, and improve the overall prediction efficiency on the basis of ensuring the accuracy of the prediction.

[0107] Step 300: Use a time series model to perform feature prediction on the corrected traffic feature vector to obtain a predicted traffic state representation vector.

[0108] In a feasible embodiment, the corrected traffic feature vector is input into a time series model for feature prediction to obtain a predicted traffic state representation vector.

[0109] Optionally, the time series model is an LSTM model.

[0110] Specifically, the long-term traffic flow evolution characteristics are extracted by controlling the forgetting unit, updating unit and feature integration unit of the LSTM model and managing the temporal impact representation vector. Among them, the forgetting unit accurately controls the degree of forgetting of historical information, the updating unit accurately screens and integrates the input information at the current moment, and the feature integration unit carefully determines the output part of the temporal impact representation vector. Through the above mechanism, the topological traffic evolution representation unit can better handle the long-term traffic flow evolution characteristics in time series data and avoid the problem of gradient disappearance or explosion. The expressions corresponding to the specific units are as follows:

[0111]

[0112] ht =o t tanh(C t )

[0113] Among them, f t is the forgetting unit at time t, controlling the last time sequence to influence the representation vector, c t-1 The degree to which the information in the t is the update unit at time t, controlling the new input The degree to which the information in is added to the traffic state representation vector; is the candidate state at time t, used to calculate the new traffic state representation vector C t , C t is the traffic status representation vector at time t, which is represented by the forgetting unit f t and update unit i t Control update, including memory information at the current moment; o t is the feature integration unit at time t, controlling the timing impact representation vector C t How to influence the traffic state representation vector h t ;h t is the traffic status representation vector at time t; W f , W i , W o are the weight matrices to be trained corresponding to the unit information; U f , U i , U o are the weight matrices to be trained corresponding to the unit traffic state representation vectors; b f 、b i 、b o are the bias items to be trained for the corresponding units respectively; W C 、b C is the weight matrix and bias term of the candidate state, σ(·) is the Sigmoid activation function, and tanh(·) is the tanh activation function.

[0114] It should also be noted that in the embodiment of the present application, the time series model can predict the traffic state characterization vector sequence corresponding to the next cycle based on the input of multiple historical cycle flow curves. Specifically, it can be expressed as:

[0115] H out =[h t’-T h t’-T+1 …h t’ ]

[0116] Among them, h t‘ is the traffic status characterization vector at time t', and time t' is the predicted time t in the next cycle.

[0117] It should be understood that since the original traffic feature vector used for prediction is determined based on the disturbance residual of the periodic traffic curve and historical traffic data, the predicted traffic state characterization vector is a prediction of the degree of change of traffic flow, that is, it is not a direct prediction of traffic flow. Therefore, further decoding and fitting are required to obtain the real predicted traffic flow information.

[0118] Step 400: Determine predicted traffic flow information corresponding to the traffic section to be predicted based on the predicted traffic state characterization vector.

[0119] In a possible embodiment, if Figure 4 As shown, step 400, based on the predicted traffic state characterizing the sales volume, determines the predicted traffic flow information corresponding to the traffic section to be predicted, including:

[0120] Step 410 , decoding the predicted traffic state representation vector based on the regression relationship between the predicted feature parameters and the predicted value traffic state, and obtaining a predicted output value corresponding to the traffic section to be predicted.

[0121] That is to say, the embodiment of the present application constructs a regression relationship between the predicted characteristic parameters and the predicted value of the traffic state, that is, by processing the characteristic parameters, applying neuron discarding and activation functions and other methods to alleviate the overfitting problem, the regression relationship between the predicted characteristic parameters and the predicted value of the traffic state is constructed. Specifically, the following expression can be used:

[0122]

[0123] in, is the predicted output value, H out is the traffic state representation vector sequence of the predicted output, W F1 , W F2 , are the weight matrices to be trained for the corresponding feature decoding units, and b F1 、b F2 are the weight matrices to be trained for the corresponding feature decoding units, dropout(·) is the neuron dropout function used to reduce overfitting, and tanh(·) is the tanh activation function.

[0124] Step 420, based on the periodic flow curve and the predicted output value, determine the predicted traffic flow information corresponding to the traffic section to be predicted.

[0125] Specifically, the following expression can be used:

[0126]

[0127] in, is the predicted traffic flow information of road section i, D i is the flow curve of the previous cycle of section i, is the predicted output value.

[0128] Preferably, when determining the predicted traffic flow information corresponding to the traffic section to be predicted, the average value of a plurality of historical period flow curves may be used to normalize the non-difference information.

[0129] To summarize, the freight flow prediction method based on the spatiotemporal evolution of the topological network proposed in the embodiment of the present application constructs the original flow feature vector using a three-dimensional traffic flow information matrix that expresses more traffic flow differences, converts traffic flow into flow difference information with a finer time granularity, realizes the prediction of the intensity of flow difference changes, improves the attention to differential flow, and at the same time avoids the edge missing caused by the overall flow prediction, thereby improving the accuracy of finer-grained flow prediction.

[0130] It should be noted that although the operations of the method of the present invention are described in a particular order in the drawings, this does not require or imply that the operations must be performed in this particular order or that all illustrated operations must be performed to achieve desired results.

[0131] Figure 5 A schematic diagram of the structure of a freight flow prediction device based on spatiotemporal evolution of a topological network provided in one embodiment of the present application is shown.

[0132] like Figure 5 As shown, the freight flow prediction device 10 based on the spatiotemporal evolution of the topological network includes:

[0133] The first determination module 11 is used to construct a topological network based on the road information of the traffic section to be predicted, and determine the original traffic feature vector corresponding to each section in the topological network according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix; the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which the section belongs;

[0134] A correction module 12, configured to correct the original flow characteristic vector for influence intensity characteristics based on the influence relationship between the road sections in the topological network, to obtain a corrected flow characteristic vector;

[0135] A prediction module 13, used to perform feature prediction on the corrected traffic feature vector using a time series model to obtain a predicted traffic state representation vector;

[0136] The second determination module 14 is used to determine the predicted traffic flow corresponding to the traffic section to be predicted based on the predicted traffic state representation vector.

[0137] In some embodiments, the first determining module 11 is further configured to:

[0138] Constructing a traffic matrix based on the historical traffic data corresponding to the traffic section to be predicted;

[0139] Determine, according to the flow matrix, a periodic flow curve corresponding to the historical flow data;

[0140] Based on the periodic flow curve and the historical flow data, construct the three-dimensional traffic flow information matrix;

[0141] Based on the three-dimensional traffic flow information matrix, the original traffic flow feature vector corresponding to each road section in the topological network is determined.

[0142] In some embodiments, the first determining module 11 is further configured to:

[0143] A three-dimensional traffic flow information matrix is ​​constructed based on the periodic flow curve and the disturbance residual of the historical flow data.

[0144] In some embodiments, the correction module 12 is further configured to:

[0145] Based on the self-matching-shared matching mechanism, calculating the influence intensity coefficient between the first-order adjacent sections in the topological network;

[0146] Based on the impact intensity coefficient, the original flow characteristic vector is corrected by impact intensity characteristics to obtain a corrected flow characteristic vector.

[0147] In some embodiments, the correction module 12 is further configured to:

[0148] Normalizing the impact intensity coefficient;

[0149] Using the normalized impact intensity coefficient, respectively correcting the impact intensity characteristics of the original flow characteristic vector;

[0150] The corrected average value of the K dimensions is used as the corrected traffic feature vector.

[0151] In some embodiments, the prediction module 13 is further configured to:

[0152] The modified traffic feature vector is input into the time series model for feature prediction to obtain a predicted traffic state representation vector.

[0153] In some embodiments, the second determining module 14 is further configured to:

[0154] Decoding the predicted traffic state representation vector based on the regression relationship between the predicted characteristic parameter and the predicted value traffic state to obtain a predicted output value corresponding to the traffic section to be predicted;

[0155] Based on the periodic flow curve and the predicted output value, the predicted traffic flow information corresponding to the traffic section to be predicted is determined.

[0156] It should be understood that the modules or modules recorded in the freight flow prediction device 10 based on the spatiotemporal evolution of the topological network are similar to those in the reference Figure 1 The various steps in the described method correspond to each other. Therefore, the operations and features described above for the method are also applicable to the freight flow prediction device 10 based on the spatiotemporal evolution of the topological network and the modules contained therein, and will not be repeated here. The freight flow prediction device 10 based on the spatiotemporal evolution of the topological network can be pre-implemented in a browser or other security application of an electronic device, or can be loaded into the browser or its security application of an electronic device by downloading or the like. The corresponding modules in the freight flow prediction device 10 based on the spatiotemporal evolution of the topological network can cooperate with the modules in the electronic device to implement the solution of the embodiment of the present application.

[0157] For the several modules or units mentioned in the above detailed description, this division is not mandatory. In fact, according to the embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. On the contrary, the features and functions of one module or unit described above can be further divided into multiple modules or units to be embodied.

[0158] Reference below Figure 6 , Figure 6 A schematic diagram of the structure of a computer system of an electronic device or server suitable for implementing an embodiment of the present application is shown.

[0159] like Figure 6 As shown, the computer system includes a central processing unit (CPU) 601, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 602 or the program loaded from the storage part 608 to the random access memory (RAM) 603. In the RAM 603, various programs and data required for the operation instructions of the system are also stored. The CPU 601, the ROM 602 and the RAM 603 are connected to each other through a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0160] The following components are connected to the I / O interface 605: an input section 606 including a keyboard, a mouse, etc.; an output section 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 608 including a hard disk, etc.; and a communication section 609 including a network interface card such as a LAN card, a modem, etc. The communication section 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the I / O interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage section 608 as needed.

[0161] In particular, according to an embodiment of the present application, the above reference flow chart Figure 2 The described process can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a computer readable medium, and the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program includes a program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the central processing unit (CPU) 601, the above-mentioned functions defined in the system of the present application are executed.

[0162] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. Computer-readable signal media may also be any computer-readable medium such as a computer-readable storage medium that can send, propagate or transmit a program for use by or in conjunction with an instruction execution system, device or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0163] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, functions and operating instructions of the system, method and computer program product according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the aforementioned module, program segment or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the box can also occur in a sequence different from that marked in the accompanying drawings. For example, the boxes represented by two connections can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operating instruction, or can be implemented with a combination of dedicated hardware and computer instructions.

[0164] The units or modules involved in the embodiments described in the present application may be implemented by software or by hardware. The units or modules described may also be arranged in a processor, for example, they may be described as: a processor including a first determination module, a correction module, a prediction module, and a second determination module. Among them, the names of these units or modules do not constitute limitations on the units or modules themselves under certain circumstances. For example, the first determination module may also be described as "building a topological network based on the road information of the traffic section to be predicted, and determining the original traffic feature vector corresponding to each section in the topological network according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix; the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which the section belongs".

[0165] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the electronic device described in the above embodiment, or may exist independently without being assembled into the electronic device. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the freight flow prediction method based on the spatiotemporal evolution of the topological network described in the present application.

[0166] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

Claims

1. A freight flow prediction method based on the spatiotemporal evolution of topological networks, characterized in that: include: A topological network is constructed based on the road information of the traffic section to be predicted, and an original traffic feature vector corresponding to each section in the topological network is determined according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on a three-dimensional traffic flow information matrix; the three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which the section belongs; Based on the influence relationship between the road sections in the topological network, the original flow characteristic vector is corrected by the influence intensity characteristic to obtain a corrected flow characteristic vector; Using a time series model to perform feature prediction on the corrected traffic feature vector to obtain a predicted traffic state representation vector; Based on the predicted traffic state characterization vector, the predicted traffic flow information corresponding to the traffic section to be predicted is determined.

2. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 1 is characterized in that: The determining, based on the historical traffic data corresponding to the traffic section to be predicted, the original traffic feature vector corresponding to each section in the topological network comprises: Constructing a traffic matrix based on the historical traffic data corresponding to the traffic section to be predicted; Determine, according to the flow matrix, a periodic flow curve corresponding to the historical flow data; Based on the periodic flow curve and the historical flow data, construct the three-dimensional traffic flow information matrix; Based on the three-dimensional traffic flow information matrix, the original traffic flow feature vector corresponding to each road section in the topological network is determined.

3. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 2 is characterized in that: Also includes: A three-dimensional traffic flow information matrix is ​​constructed based on the periodic flow curve and the disturbance residual of the historical flow data.

4. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 1 is characterized in that: The method of performing influence intensity feature correction on the original flow feature vector based on the influence relationship between the road sections in the topological network to obtain a corrected flow feature vector includes: Based on the self-matching-shared matching mechanism, calculating the influence intensity coefficient between the first-order adjacent sections in the topological network; Based on the impact intensity coefficient, the original flow characteristic vector is corrected by impact intensity characteristics to obtain a corrected flow characteristic vector.

5. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 4 is characterized in that: The step of performing influence intensity characteristic correction on the original flow characteristic vector based on the influence intensity coefficient to obtain a corrected flow characteristic vector includes: Normalizing the impact intensity coefficient; Using the normalized impact intensity coefficient, respectively correcting the impact intensity characteristics of the original flow characteristic vector; The corrected average value of the K dimensions is used as the corrected traffic feature vector.

6. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 1 is characterized in that: The method of using a time series model to perform feature prediction on the corrected traffic feature vector to obtain a predicted traffic state representation vector includes: The modified traffic feature vector is input into the time series model for feature prediction to obtain a predicted traffic state representation vector.

7. The freight flow prediction method based on the spatiotemporal evolution of topological networks according to claim 1 is characterized in that: The step of determining predicted traffic flow information corresponding to the traffic section to be predicted based on the predicted traffic state representation vector includes: Decoding the predicted traffic state representation vector based on the regression relationship between the predicted characteristic parameter and the predicted value traffic state to obtain a predicted output value corresponding to the traffic section to be predicted; Based on the periodic flow curve and the predicted output value, the predicted traffic flow information corresponding to the traffic section to be predicted is determined.

8. A freight flow prediction device based on the spatiotemporal evolution of a topological network, characterized in that: include: A first determination module is used to construct a topological network based on the road information of the traffic section to be predicted, and determine the original traffic feature vector corresponding to each section in the topological network according to the historical traffic data corresponding to the traffic section to be predicted; the original traffic feature vector is an information flow time series constructed based on the three-dimensional traffic flow information matrix; The three-dimensional traffic flow information matrix is ​​related to a first time dimension, a second time dimension within the first time dimension, and a spatial dimension to which a road segment belongs; A correction module, used for correcting the influence intensity characteristics of the original flow characteristic vector based on the influence relationship between the road sections in the topological network to obtain a corrected flow characteristic vector; A prediction module, used for performing feature prediction on the corrected traffic feature vector using a time series model to obtain a predicted traffic state representation vector; The second determination module is used to determine the predicted traffic flow corresponding to the traffic section to be predicted based on the predicted traffic state representation vector.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, it implements the freight flow prediction method based on the spatiotemporal evolution of the topological network as described in any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, it implements the freight flow prediction method based on the spatiotemporal evolution of the topological network as described in any one of claims 1 to 7.

Images