A hotspot path identification method and related equipment

Abstract

The present invention provides a hotspot path identification method and related equipment, comprising: obtaining vehicle operation data collected by multiple roadside sensors, and obtaining the travel time and initial trajectory of a target vehicle based on the vehicle operation data; calculating a travel threshold by travel time to divide the initial trajectory into multiple path trajectories; constructing a topological road network including edges and network nodes based on the multiple path trajectories and multiple roadside sensors; inputting the vehicle operation data and the topological road network into a snake clustering model for clustering to obtain a membership matrix including multiple travel modes and multiple key nodes; using the membership matrix to correct the number of vehicles passing each edge to obtain a correction result, constructing a Steiner tree for representing a set of paths connecting all key points under the multiple travel modes, and identifying hotspot paths; and realizing hotspot path identification for vehicle passing data with relatively sparse sensor distribution.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation technology, and in particular to a hotspot path identification method and related equipment. Background Art

[0002] Hotspot paths refer to routes where vehicles frequently pass within a certain period of time. They can better reflect vehicle movement patterns and travel rules, and can further reflect the traffic conditions of urban road networks. They are of great significance in travel induction, commercial site selection and traffic management.

[0003] As the intelligent connected vehicle industry and the development of vehicle-road cooperative systems gain increasing attention, the deployment density of intelligent connected vehicle equipment is gradually increasing. Intelligent connected vehicle-road systems can collect comprehensive vehicle operation data, and related research is gaining increasing attention. Intelligent connected vehicle operation data is traffic data acquired by intelligent connected roadside sensors, including information such as license plate numbers, vehicle speeds, and acquisition times. It can effectively record the spatiotemporal evolution of vehicle traffic within the road network. Intelligent connected vehicle operation data offers advantages such as high accuracy, comprehensiveness, and large data volumes. However, the models and methods proposed by existing research are not well suited for identifying hotspot paths based on intelligent connected vehicle operation data.

[0004] Based on different research objects, relevant researchers have divided hotspot path research into hotspot path research based on trajectory data and hotspot path research based on road network traffic attributes. Among them, hotspot path research based on trajectory data mainly measures trajectory similarity through the traffic information of the trajectory data itself or indicators after data processing, thereby clustering trajectories to achieve hotspot path identification; however, research based on trajectory data requires the use of vehicle trajectory data composed of continuous trajectory points for analysis, and requires the data to have a low sampling time interval, which is not suitable for detecting vehicle operation data with fixed positions. Hotspot path research based on road network traffic attributes mainly obtains road network traffic attributes through traffic data and analyzes them in combination with the road network topology to achieve hotspot path identification; however, research based on road network traffic attributes cannot perform calculations when the road network topology is incomplete. It is suitable for data with dense distribution on the road network. When the distribution of roadside sensors is relatively sparse, it is difficult to obtain reliable processing results. Summary of the Invention

[0005] The present invention provides a hotspot path identification method and related equipment, which aims to solve the problem of identifying hotspot paths for vehicle passing data with relatively sparse sensor distribution.

[0006] In order to achieve the above object, the present invention provides a hotspot path identification method, comprising:

[0007] Step 1: Obtain vehicle operation data collected by multiple roadside sensors, the vehicle operation data including the target vehicle's license plate information, time information, and latitude and longitude information, and obtain the target vehicle's travel time and initial trajectory based on the vehicle operation data;

[0008] Step 2: Calculate the travel threshold according to the travel time, divide the initial trajectory into multiple paths by the travel threshold, and obtain multiple paths.

[0009] Step 3: Input the set of multiple path trajectories as a corpus, the multiple roadside sensors as words, and each path trajectory as a sentence into the word vector model for training. Multiple word vectors corresponding to the roadside sensors are obtained, and the word vectors are used as network nodes of the topological road network.

[0010] Step 4: For any two word vectors from the multiple word vectors, calculate the cosine distance between any two word vectors, use the path between two roadside sensors whose cosine distance is lower than the correlation threshold as an edge of the topological road network, and construct a topological road network including edges and network nodes;

[0011] Step 5: Input the vehicle operation data and the topological road network into the snake clustering model for clustering to obtain a membership matrix, and then obtain the travel mode and key nodes according to the membership matrix;

[0012] Step 6: Use the membership matrix to correct the number of vehicles passing each edge in the topological road network to obtain the correction result. Take the inverse of the correction result as the weight to construct a Steiner tree for representing the set of paths connecting all key points under multiple travel modes. In the Steiner tree, identify the path with a number of vehicles passing greater than the preset threshold as the hotspot path.

[0013] More specifically, step 1 includes:

[0014] Using vehicle operation data, the spatial location distribution points of the target vehicle at different times are connected in series to form the initial trajectory;

[0015] The initial trajectory is filtered by the upper and lower limits of the target vehicle's travel time to obtain the valid trajectory of the target vehicle;

[0016] Use the mean and median values to clean up outliers on the valid trajectory and obtain the cleaning results;

[0017] The travel time of the target vehicle between roadside sensors is calculated using the cleaned results.

[0018] Furthermore, the travel threshold is calculated by the travel time. The expression of the travel threshold is:

[0019] t thr =t max +t0

[0020] Among them, t thr represents the travel threshold, t max It represents the upper limit of travel time, and t0 represents the maximum possible duration of temporary parking.

[0021] More specifically, step 4 includes:

[0022] By formula Calculate the cosine distance between any two word vectors, where d represents the cosine distance, and A and B represent the word vectors corresponding to two adjacent roadside sensors;

[0023] Set the correlation threshold θ between two adjacent roadside sensors;

[0024] When the cosine distance is lower than the correlation threshold, the path between the two roadside sensors corresponding to the cosine distance is taken as the edge of the topological road network, and the cosine distance is used as the weight of the edge;

[0025] A topological road network is constructed based on the edges of the topological road network and the network nodes of the topological road network.

[0026] Furthermore, the vehicle operation data and the topological road network are input into the snake clustering model for clustering, and the membership matrix is obtained, including:

[0027] The average passing speed of each roadside sensor is obtained through vehicle operation data. The average passing speed is used to characterize the traffic attributes of the network node.

[0028] The average passing speed and topological road network of each roadside sensor are input into the snake clustering model for reconstruction to obtain the network weight matrix;

[0029] For each network node in the topological road network, a snake list is generated for each network node, and the similarity between any two network nodes is calculated based on the snake list;

[0030] According to the similarity, the network weight matrix is clustered using the symmetric non-negative matrix factorization method to obtain the membership matrix.

[0031] Specifically, the similarity between any two network nodes is:

[0032]

[0033] Among them, w(i,j) represents the similarity between network node i and network node j, N represents the maximum length of the snake list, represents the sorting weight coefficient in the snake list, intersect represents the number of common elements in the two sublists, S ik represents the sublist consisting of the first k elements in the snake list corresponding to the network node i, S jkRepresents the sublist consisting of the first k elements in the snake list corresponding to network node j.

[0034] The present invention also provides a hotspot path identification device, comprising:

[0035] An acquisition module is used to obtain vehicle operation data collected by multiple roadside sensors. The vehicle operation data includes the license plate information, time information, and latitude and longitude information of the target vehicle, and obtain the travel time and initial trajectory of the target vehicle based on the vehicle operation data;

[0036] A partitioning module is used to calculate the travel threshold according to the travel time, and to partition the initial trajectory by the travel threshold to obtain multiple path trajectories;

[0037] A training module is used to input a set of multiple path trajectories as a corpus, multiple roadside sensors as words, and each path trajectory as a sentence into a word vector model for training, thereby obtaining multiple word vectors corresponding to the roadside sensors and using the word vectors as network nodes of the topological road network;

[0038] A construction module is configured to calculate a cosine distance between any two word vectors from the plurality of word vectors, use a path between two roadside sensors whose cosine distance is lower than a correlation threshold as an edge of a topological road network, and construct a topological road network including edges and network nodes;

[0039] The clustering module is used to input vehicle operation data and topological road network into the snake clustering model for clustering to obtain a membership matrix. Based on the membership matrix, multiple travel modes and multiple key nodes are obtained;

[0040] The identification module is used to use the membership matrix to correct the number of vehicles passing each edge in the topological road network, obtain the correction result, take the inverse of the correction result as the weight, and construct a Steiner tree for representing the set of paths connecting all key points under multiple travel modes. In the Steiner tree, the path with a number of vehicles passing greater than a preset threshold is identified as the hot path.

[0041] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the hotspot path identification method is implemented.

[0042] The present invention also provides a terminal device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the hotspot path identification method is implemented.

[0043] The above solution of the present invention has the following beneficial effects:

[0044] The present invention obtains vehicle operation data collected by multiple roadside sensors and, based on the vehicle operation data, obtains the travel time and initial trajectory of a target vehicle. The initial trajectory is then divided into multiple path trajectories based on the travel time and a travel threshold. A topological road network comprising edges and network nodes is constructed based on the multiple path trajectories and multiple roadside sensors. The vehicle operation data and the topological road network are clustered using a snake clustering model to obtain a membership matrix comprising multiple travel modes and multiple key nodes. The membership matrix is used to modify the number of vehicles passing each edge, obtaining a modified result. A Steiner tree is constructed to represent the set of paths connecting all key points under the multiple travel modes and to identify hotspot paths. Compared to existing technologies, the present invention converts vehicle passing data into travel trajectories and reconstructs the road network using a word vector model, thus overcoming the difficulty in obtaining a topological road network based on the actual road network when sensors are sparse. Furthermore, sensors are used as topological road network nodes, and the attributes of the topological road network nodes are combined with the network structure to identify hotspot paths using a snake clustering algorithm and a Steiner tree. This enables hotspot path identification for vehicle passing data with relatively sparse sensor distribution.

[0045] Other beneficial effects of the present invention will be described in detail in the subsequent specific implementation section. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of a flow chart of an embodiment of the present invention;

[0047] Figure 2 Schematic diagram of the structure of the skip-gram model in an embodiment of the present invention;

[0048] Figure 3 A schematic diagram of a topological road network constructed in an embodiment of the present invention;

[0049] Figure 4 SNMF clustering diagram in an embodiment of the present invention;

[0050] Figure 5 Schematic diagram of the Steiner tree in an embodiment of the present invention. DETAILED DESCRIPTION

[0051] To make the technical problems, technical solutions, and advantages to be solved by the present invention more clear, the following is a detailed description with reference to the accompanying drawings and specific embodiments. It is obvious that the embodiments described are only some of the embodiments of the present invention, not all of them. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0052] In the description of the present invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings and are intended solely to facilitate and simplify the description of the present invention. They are not intended to indicate or imply that the devices or components referred to must have, be constructed, or operate in a specific orientation, and therefore should not be construed as limitations on the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0053] In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood broadly. For example, they may refer to a locking connection, a detachable connection, or an integral connection; they may refer to a mechanical connection or an electrical connection; they may refer to a direct connection or an indirect connection through an intermediate medium; and they may refer to internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on the specific circumstances.

[0054] In addition, the technical features involved in the different embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0055] In view of the existing problems, the present invention provides a hotspot path identification method and related equipment.

[0056] like Figure 1 As shown, an embodiment of the present invention provides a hotspot path identification method, comprising:

[0057] Step 1: Obtain vehicle operation data collected by multiple roadside sensors, the vehicle operation data including the target vehicle's license plate information, time information, and latitude and longitude information, and obtain the target vehicle's travel time and initial trajectory based on the vehicle operation data;

[0058] Step 2: Calculate the travel threshold according to the travel time, divide the initial trajectory into multiple paths by the travel threshold, and obtain multiple paths;

[0059] Step 3: Input the set of multiple path trajectories as a corpus, the multiple roadside sensors as words, and each path trajectory as a sentence into the word vector model for training. Multiple word vectors corresponding to the roadside sensors are obtained, and the word vectors are used as network nodes of the topological road network.

[0060] Step 4: For any two word vectors from the multiple word vectors, calculate the cosine distance between any two word vectors, use the path between two roadside sensors whose cosine distance is lower than the correlation threshold as an edge of the topological road network, and construct a topological road network including edges and network nodes;

[0061] Step 5: Input the vehicle operation data and the topological road network into the snake clustering model for clustering to obtain a membership matrix, and then obtain the travel mode and key nodes according to the membership matrix;

[0062] Step 6: Use the membership matrix to correct the number of vehicles passing each edge in the topological road network to obtain the correction result. Take the inverse of the correction result as the weight to construct a Steiner tree for representing the set of paths connecting all key points under multiple travel modes. In the Steiner tree, identify the path with a number of vehicles passing greater than the preset threshold as the hotspot path.

[0063] Specifically, step 1 includes:

[0064] Using vehicle operation data, the spatial location distribution points of the target vehicle at different times are connected in series to form the initial trajectory;

[0065] The initial trajectory is filtered by the upper and lower limits of the target vehicle's travel time to obtain the valid trajectory of the target vehicle;

[0066] Use the mean and median values to clean up outliers on the valid trajectory and obtain the cleaning results;

[0067] The travel time of the target vehicle between roadside sensors is calculated using the cleaned results.

[0068] In this embodiment of the present invention, data cleansing should be performed before trip segmentation to prevent outliers in travel time from affecting trip thresholds. The acquired data is the full-time travel time data between roadside sensors, and the minimum processing unit is the travel time data within a single time window. This prevents inaccurate outlier detection results due to the time-varying nature of travel time. The time window setting is affected by data characteristics. Setting the time window too large can easily cause travel time variations to affect the results, while setting the time window too small can easily result in too little data being processed, which can also affect the results. While vehicle operation data has high reliability, false detections and missed detections may still occur. Therefore, vehicle operation data should be filtered using upper and lower travel time limits. The values of these limits are determined by the theoretical travel time between roadside sensors. Vehicles that have been stationary for a long time and then re-enter the driving state may generate data above the upper travel time limit, while vehicles with fake license plates or speeding may generate data below the lower travel time limit. After obtaining valid trajectories of the target vehicle using the upper and lower travel time limits, outliers are removed using the mean and median values to calculate the travel time between roadside sensors.

[0069] Specifically, the travel threshold is calculated by the travel time, and the expression of the travel threshold is:

[0070] t thr =t max +t0

[0071] Among them, t thr represents the travel threshold, t max It represents the upper limit of travel time, and t0 represents the maximum possible duration of temporary parking.

[0072] The embodiment of the present invention can connect the spatial position distribution points of the target vehicle at different times into an initial trajectory by using the target vehicle's license plate information, time information, and latitude and longitude information, and use the spatial position distribution points where the travel time is higher than the travel threshold as stop points to obtain multiple path trajectories.

[0073] For vehicle operation data, a topological road network cannot be directly generated from the actual road network. In a topological road network directly converted from the actual road network, the network nodes are road intersections. However, in actual application scenarios, the location distribution of roadside sensors and road intersections rarely completely overlap. There is a high probability that there will be intersections without roadside sensors or roadside sensors deployed at non-intersection locations. Therefore, it is necessary to construct a topological road network with roadside sensors as network nodes by using the path trajectories obtained from multiple trips.

[0074] A word vector model is an unsupervised model trained on a text corpus. It converts words into word vectors and calculates their coordinates in a high-dimensional space, thereby measuring semantic similarity between words. This embodiment of the present invention uses the skip-gram model within the word vector model. This model uses a dataset of multiple road trajectory data as a corpus, multiple roadside sensors as words, and each road trajectory as a sentence input for word vector training. This model generates multiple word vectors corresponding to the roadside sensors, achieving the conversion of roadside sensors into word vectors to quantitatively evaluate the correlation between sensors.

[0075] like Figure 2 As shown in the figure, the skip-gram model is a network model consisting of an input layer, a hidden layer, and an output layer. t The input layer predicts the context and trains the network weights to make the network output closer to the target word w t After training, the network weights of the context words are the word vectors. The objective function P of the skip-gram model is as follows:

[0076]

[0077] Among them, k represents the number of adjacent words in the context, p(w t+i |w t ) indicates that according to the target word w t Correctly predict adjacent words w t+i The probability of , T represents the total number of words in the corpus.

[0078] After obtaining the word vectors corresponding to the roadside sensors through skip-gram model training, the correlation between sensors is judged by cosine distance, and the formula Calculate the cosine distance between any two word vectors, where d represents the cosine distance, and A and B represent the word vectors corresponding to two adjacent roadside sensors;

[0079] Set the correlation threshold θ between two adjacent roadside sensors;

[0080] Based on the calculated cosine distance and the correlation threshold θ, when the cosine distance is lower than the correlation threshold, the path between the two roadside sensors corresponding to the cosine distance is used as the edge of the topological road network, and the cosine distance is used as the weight of the edge; based on the edges and network nodes, a topological road network is constructed, such as Figure 3 shown.

[0081] Specifically, the vehicle operation data and topological road network are input into the snake clustering model for clustering, and the membership matrix is obtained, including:

[0082] The average passing speed of each roadside sensor is obtained through vehicle operation data. The average passing speed is used to characterize the traffic attributes of the network node.

[0083] The average passing speed and topological road network of each roadside sensor are input into the snake clustering model and reconstructed to obtain the network weight matrix;

[0084] For each network node in the topological road network, a snake list is generated for each network node, and the similarity between any two network nodes is calculated based on the snake list;

[0085] According to the similarity, the network weight matrix is clustered using the symmetric non-negative matrix factorization method to obtain the membership matrix.

[0086] The snake clustering model is a clustering model that takes into account both network structure and node attributes. It reconstructs a network weight matrix based on the network topology and node attributes, and then clusters the reconstructed network weight matrix to discover connections between network nodes. The snake clustering model first generates a corresponding snake for each network node based on the topology and node attributes. It then calculates the similarity between network nodes based on the corresponding snakes, and finally clusters the network nodes based on the similarity between the nodes.

[0087] The specific steps for generating the corresponding snake for each node are as follows:

[0088] (1) Add the target network node to the empty list L as the initial value of the "snake";

[0089] (2) Obtain all adjacent nodes of any network node in the list to form a set A;

[0090] (3) Select a node from set A and add it to list L so that the variance of list L is minimized;

[0091] (4) Update set A. If set A is not an empty set, return to step (2); otherwise, end the loop.

[0092] The expression for calculating the similarity between network nodes based on the corresponding snake generated by the network nodes is as follows:

[0093]

[0094] Among them, w(i,j) represents the similarity between network node i and network node j, N represents the maximum length of the snake list, represents the sorting weight coefficient in the snake list, intersect represents the number of common elements in the two sublists, S ik represents the sublist consisting of the first k elements in the snake list corresponding to the network node i, S jk Represents the sublist consisting of the first k elements in the snake list corresponding to network node j.

[0095] After obtaining the node similarity w(i,j), clustering is performed using the Symmetric Non-negative Matrix Factorization (SNMF) method, as follows: Figure 4 As shown, SNMF is a low-rank matrix approximation method, and its objective function F is as follows:

[0096]

[0097] in, represents the normalized matrix W, and H is the membership matrix.

[0098] The embodiment of the present invention satisfies the following through approximate calculation: The membership matrix is calculated based on the matrix of the skip-gram model; based on the topological road network obtained by training, the average speed of the sensor passing the vehicle is used to represent the traffic attributes of the network nodes. The membership of each roadside sensor with respect to multiple travel modes can be calculated according to the snake clustering algorithm, thereby obtaining multiple travel modes, and the key nodes can be determined according to the membership of the network nodes in different travel modes.

[0099] For the weighted road network topology G = (V, E) and key nodes The minimum Steiner tree is defined as the tree with the minimum weight connecting all key nodes in the graph G. Compared with the minimum spanning tree, the minimum Steiner tree allows connecting other network nodes in the weighted road network topology graph G outside a given key node R, so that the total length of the edge set that makes up the Steiner tree is minimized.

[0100] In the process of constructing the Steiner tree, the set of roadside sensors is the point set of the network graph. The roadside sensors with a membership degree higher than the preset threshold in the corresponding travel mode are the key points R, and the preset threshold is 0.5. The virtual paths between the roadside sensors in the topological road network are the edge sets of the network graph, such as Figure 5 As shown, the number of vehicles passing between roadside sensors is corrected using membership, resulting in a correction result. The inverse of the correction result is used as the weight of the edge set to construct a Steiner tree. In this embodiment of the present invention, the practical significance of the Steiner tree is a set of virtual paths between roadside sensors that have the highest possible flow rate after correction and connect all key points in the corresponding travel mode. This can represent the backbone road network for the corresponding travel mode, where the edges with the highest flow rate in the correction result are hotspot paths.

[0101] The present invention obtains vehicle operation data collected by multiple roadside sensors and, based on the vehicle operation data, obtains the travel time and initial trajectory of a target vehicle. The initial trajectory is divided into multiple path trajectories by calculating a travel threshold based on the travel time. A topological road network comprising edges and network nodes is constructed based on the multiple path trajectories and multiple roadside sensors. The vehicle operation data and the topological road network are clustered using a snake clustering model to obtain a membership matrix comprising multiple travel modes and multiple key nodes. The membership matrix is used to modify the number of vehicles passing each edge to obtain a modified result. A Steiner tree is constructed to represent a set of paths connecting all key points under the multiple travel modes and to identify hotspot paths. Compared to existing technologies, the present invention converts vehicle passing data into travel trajectories and reconstructs the road network using a word vector model, thus overcoming the difficulty in obtaining a topological road network based on the actual road network when sensors are sparse. Furthermore, sensors are used as topological road network nodes. The snake clustering algorithm and the Steiner tree are combined with the topological road network node attributes and network structure to identify hotspot paths, thereby enabling hotspot path identification for vehicle passing data with relatively sparse sensor distribution.

[0102] An embodiment of the present invention further provides a hotspot path identification device, comprising:

[0103] An acquisition module is used to obtain vehicle operation data collected by multiple roadside sensors. The vehicle operation data includes the license plate information, time information, and latitude and longitude information of the target vehicle, and obtain the travel time and initial trajectory of the target vehicle based on the vehicle operation data;

[0104] A partitioning module is used to calculate the travel threshold according to the travel time, and to partition the initial trajectory by the travel threshold to obtain multiple path trajectories;

[0105] A training module is used to input a set of multiple path trajectories as a corpus, multiple roadside sensors as words, and each path trajectory as a sentence into a word vector model for training, thereby obtaining multiple word vectors corresponding to the roadside sensors and using the word vectors as network nodes of the topological road network;

[0106] A construction module is configured to calculate a cosine distance between any two word vectors from the plurality of word vectors, use a path between two roadside sensors whose cosine distance is lower than a correlation threshold as an edge of a topological road network, and construct a topological road network including edges and network nodes;

[0107] The clustering module is used to input vehicle operation data and topological road network into the snake clustering model for clustering to obtain a membership matrix. Based on the membership matrix, multiple travel modes and multiple key nodes are obtained;

[0108] The identification module is used to use the membership matrix to correct the number of vehicles passing each edge in the topological road network, obtain the correction result, take the inverse of the correction result as the weight, and construct a Steiner tree for representing the set of paths connecting all key points under multiple travel modes. In the Steiner tree, the path with a number of vehicles passing greater than a preset threshold is identified as the hot path.

[0109] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiment of the embodiment of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0110] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the embodiments of the present invention. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0111] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the hotspot path identification method is implemented.

[0112] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the embodiments of the present invention implement all or part of the processes in the above-mentioned embodiment method, which can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and when the computer program is executed by the processor, it can implement the steps of the above-mentioned various method embodiments. Wherein, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include at least: any entity or device capable of carrying the computer program code to a construction device / terminal device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electric carrier signal, a telecommunication signal and a software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, a computer-readable medium cannot be an electric carrier signal or a telecommunication signal.

[0113] An embodiment of the present invention further provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements a hotspot path identification method when executing the computer program.

[0114] It should be noted that the terminal device may be a mobile phone, tablet computer, laptop computer, ultra-mobile personal computer (UmPC), netbook, personal digital assistant (PDA), and other terminal devices. For example, the terminal device may be a station (ST, STAiON) in a WLAN, a cellular phone, a cordless phone, a Session Initiation Protocol (SiP) phone, a Wireless Local Loop (WLL) station, a Personal Digital Assistant (PDA), a handheld device with wireless communication capabilities, a computing device or other processing device connected to a wireless modem, a computer, a laptop computer, a handheld communication device, a handheld computing device, a satellite wireless device, etc. The embodiments of the present invention do not impose any restrictions on the specific type of the terminal device.

[0115] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASiC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0116] In some embodiments, the memory may be an internal storage unit of the terminal device, such as a hard disk or memory of the terminal device. In other embodiments, the memory may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device, a smart memory card (SmC, Smart media Card), a secure digital (SD, Secure Digital) card, a flash card, etc. Furthermore, the memory may also include both an internal storage unit of the terminal device and an external storage device. The memory is used to store an operating system, an application program, a boot loader (BootLoader), data, and other programs, such as the program code of the computer program. The memory may also be used to temporarily store data that has been output or is to be output.

[0117] It should be noted that the information interaction, execution process, etc. between the above-mentioned devices / units are based on the same concept as the method embodiments of the embodiments of the present invention. Their specific functions and technical effects can be found in the method embodiment section and will not be repeated here.

[0118] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as within the scope of protection of the present invention.

Claims

1. A hotspot path identification method, characterized in that: include: Step 1: Acquire vehicle operation data collected by multiple roadside sensors, wherein the vehicle operation data includes license plate information, time information, and latitude and longitude information of the target vehicle, and obtain the travel time and initial trajectory of the target vehicle based on the vehicle operation data; Step 2: Calculate a travel threshold according to the travel time, and divide the initial trajectory into multiple paths according to the travel threshold to obtain multiple paths; Step 3: Input a set of the plurality of path trajectories as a corpus, the plurality of roadside sensors as words, and each path trajectory as a sentence into a word vector model for training, thereby obtaining a plurality of word vectors corresponding to the roadside sensors, and using the word vectors as network nodes of a topological road network; Step 4: for any two word vectors among the plurality of word vectors, calculating the cosine distance between any two word vectors, taking the path between the two roadside sensors whose cosine distance is lower than a correlation threshold as an edge of a topological road network, and constructing a topological road network including edges and network nodes; Step 5: Input the vehicle operation data and the topological road network into a snake clustering model for clustering to obtain a membership matrix, and obtain multiple travel modes and multiple key nodes according to the membership matrix; Step 6: Use the membership matrix to correct the number of vehicles passing each edge in the topological road network to obtain a correction result, take the inverse of the correction result as a weight, construct a Steiner tree for representing a set of paths connecting all key points under multiple travel modes, and identify in the Steiner tree a path with a number of vehicles passing greater than a preset threshold as a hotspot path.

2. The hotspot path identification method according to claim 1, characterized in that: The step 1 comprises: Using the vehicle operation data, connecting the spatial position distribution points of the target vehicle at different times into an initial trajectory; The initial trajectory is filtered by the upper and lower limits of the travel time of the target vehicle to obtain a valid trajectory of the target vehicle; Cleaning outliers on the valid trajectory using the mean and median to obtain a cleaning result; The travel time of the target vehicle between the roadside sensors is calculated based on the cleaning results.

3. The hotspot path identification method according to claim 2, characterized in that: The travel threshold is calculated by the travel time, and the expression of the travel threshold is: t thr =t max +t0 Among them, t thr represents the travel threshold, t max It represents the upper limit of travel time, and t0 represents the maximum possible duration of temporary parking.

4. The hotspot path identification method according to claim 1, characterized in that: The step 4 comprises: By formula Calculate the cosine distance between any two of the word vectors, where d represents the cosine distance, and A and B represent the word vectors corresponding to two adjacent roadside sensors; Setting a correlation threshold θ between two adjacent roadside sensors; When the cosine distance is lower than the correlation threshold, the path between the two roadside sensors corresponding to the cosine distance is used as an edge of the topological road network, and the cosine distance is used as the weight of the edge; A topological road network is constructed according to the edges of the topological road network and the network nodes of the topological road network.

5. The hotspot path identification method according to claim 4, characterized in that: The vehicle operation data and the topological road network are input into the snake clustering model for clustering to obtain a membership matrix, including: Obtaining an average vehicle passing speed of each roadside sensor through the vehicle operation data, wherein the average vehicle passing speed is used to characterize traffic attributes of the network node; The average passing speed of each roadside sensor and the topological road network are input into the snake clustering model for reconstruction to obtain a network weight matrix; For each network node in the topological road network, a snake list is generated for each network node, and similarity between any two network nodes is calculated based on the snake list; According to the similarity, the network weight matrix is clustered using a symmetric non-negative matrix decomposition method to obtain a membership matrix.

6. The hotspot path identification method according to claim 5, characterized in that: The similarity between any two network nodes is: Among them, w(i,j) represents the similarity between network node i and network node j, N represents the maximum length of the snake list, represents the sorting weight coefficient in the snake list, intersect represents the number of common elements in the two sublists, S ik represents the sublist consisting of the first k elements in the snake list corresponding to the network node i, S jk Represents the sublist consisting of the first k elements in the snake list corresponding to network node j.

7. A hotspot path identification device, characterized in that: include: an acquisition module, configured to acquire vehicle operation data collected by a plurality of roadside sensors, the vehicle operation data including license plate information, time information, and latitude and longitude information of a target vehicle, and to acquire the travel time and initial trajectory of the target vehicle based on the vehicle operation data; a partitioning module, configured to calculate a travel threshold according to the travel time, and partition the initial trajectory according to the travel threshold to obtain multiple path trajectories; a training module, configured to input a set of the plurality of path trajectories as a corpus, the plurality of roadside sensors as words, and each of the path trajectories as a sentence into a word vector model for training, thereby obtaining a plurality of word vectors corresponding to the roadside sensors, and using the word vectors as network nodes of a topological road network; a construction module, configured to calculate, for any two word vectors among the plurality of word vectors, a cosine distance between any two of the word vectors, and use a path between two roadside sensors whose cosine distance is lower than a correlation threshold as an edge of a topological road network, thereby constructing a topological road network including edges and network nodes; A clustering module is used to input the vehicle operation data and the topological road network into a snake clustering model for clustering to obtain a membership matrix, and obtain multiple travel modes and multiple key nodes according to the membership matrix; An identification module is used to use the membership matrix to correct the number of vehicles passing each edge in the topological road network to obtain a correction result, take the inverse of the correction result as a weight, and construct a Steiner tree for representing a set of paths connecting all key points under multiple travel modes. In the Steiner tree, a path with a number of vehicles passing greater than a preset threshold is identified as a hotspot path.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the hotspot path identification method according to any one of claims 1 to 6 is implemented.

9. A terminal 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 computer program, the hotspot path identification method according to any one of claims 1 to 6 is implemented.

Images