CN113255951A - Method and device for generating movement track - Google Patents
Method and device for generating movement track Download PDFInfo
- Publication number
- CN113255951A CN113255951A CN202010086899.9A CN202010086899A CN113255951A CN 113255951 A CN113255951 A CN 113255951A CN 202010086899 A CN202010086899 A CN 202010086899A CN 113255951 A CN113255951 A CN 113255951A
- Authority
- CN
- China
- Prior art keywords
- track
- generating
- historical
- virtual
- generator
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/049—Temporal neural networks, e.g. delay elements, oscillating neurons or pulsed inputs
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Business, Economics & Management (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention provides a method and a device for generating a moving track. In addition, a track quality evaluation system is constructed based on a reinforcement learning and confrontation generation network and a movement rule is synthesized, and a generation model capable of generating a more real movement track can be obtained by guiding model training and updating through automatically comparing the difference between the generation track and the real track.
Description
Technical Field
The invention relates to the technical field of virtual simulation, in particular to a method and a device for generating an individual movement track.
Background
The simulation of the movement behaviors of individuals and groups in cities has important significance for traffic control. In addition, the movement behaviors can also be applied to a city simulator and used as an underlying driving model to generate relevant tracks.
At present, the simulation model of the individual movement trajectory in the prior art is generally a model based on simple assumptions and artificial rules, for example, please refer to chinese patent application CN102393928A, a traffic simulation integration system based on interactive use of macro, medium and micro traffic simulation platforms, and the disclosed simulation system thereof includes: the system comprises a macroscopic traffic simulation module, a mesoscopic traffic simulation module, a microscopic traffic simulation module and a data acquisition and fusion module, wherein the mesoscopic traffic simulation module is connected with the macroscopic traffic simulation module, the microscopic traffic simulation module is connected with the mesoscopic traffic simulation module, and the data acquisition and fusion module is respectively connected with the macroscopic traffic simulation module, the mesoscopic traffic simulation module and the microscopic traffic simulation module. The system realizes the interactive utilization of data of each layer of simulation system and different simulation platforms on the basis of independent simulation of macroscopic, mesoscopic and microscopic traffic simulation platforms, can systematically and comprehensively simulate the traffic operation conditions of traffic flows and pedestrian flows of large-scale activities, and provides technical basis for making various traffic decision schemes of the large-scale activities.
The prior art including the above schemes mostly describes the movement behavior of an individual by several human-specified key parameters, such as defining two special geographic positions of the individual's home and place of work, and then finally outputting a simulation track by a Markov (Markov) model through transition between the home and place of work and assistance of randomly sampling other geographic positions.
The inventor finds that the prior art generally has the following problems when carrying out individual trajectory simulation:
1) based on a small amount of parameters and a simple model, the model has poor expression capability, and the generated individual movement track is simple and difficult to simulate real and complex movement behaviors such as complex time-varying and high-order transfer rules.
2) And reasonable individual movement track evaluation indexes are lacked, the simulation result can only be evaluated through a plurality of statistical indexes, parameters are manually adjusted according to the result, and the iteration efficiency is low.
Disclosure of Invention
The technical problem to be solved by the embodiments of the present invention is to provide a method and an apparatus for generating a movement trajectory, which can simulate a real auxiliary movement behavior, and guide model training and updating by comparing a difference between a generated trajectory and a real trajectory automatically, so as to obtain a model capable of generating a more real movement trajectory.
In order to solve the above technical problem, according to an aspect of the present invention, there is provided a method for generating a movement trajectory, including:
acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
inputting the data of the historical track into a generator, predicting a next position point in the historical track through the generator, and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
In accordance with at least one embodiment of the present invention, the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
According to at least one embodiment of the invention, the data of the historical track is track data within a preset time length range; generating a feature vector sequence for characterizing each historical track according to the historical track data, wherein the step comprises the following steps:
sampling a preset number of position points from data of a historical track of movement of each individual, wherein a preset time length is arranged between every two adjacent position points;
dividing the target area into a plurality of grids, and determining the grid to which each position point belongs according to the geographic position of each position point as the spatial attribute of the position point; dividing the preset time range into a plurality of time intervals, and obtaining the time interval of each position point according to the time information of each position point as the time attribute of the position point;
and generating a feature vector of each position point according to the spatial attribute and the time attribute of each position point to obtain a feature vector sequence of each historical track.
According to at least one embodiment of the invention, the data of the historical track further comprises at least one of the following information: POI information of interest points in the area where the geographic position is located, and context information of the geographic position.
According to at least one embodiment of the present invention, the preset convergence condition is: the success rate of distinguishing the virtual track by the discriminator is less than a preset threshold.
According to at least one embodiment of the present invention, the step of obtaining a determination result by determining the virtual trajectory and the sampled historical trajectory by using the determiner includes:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
According to at least one embodiment of the invention, the starting point of the simulation trajectory is generated as follows:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
According to at least one embodiment of the invention, the step of generating a virtual trajectory using the generator obtained by pre-training comprises:
and generating the rest position points of the virtual track in a random sampling mode after generating part position points of the virtual track by using the generator obtained by pre-training to obtain the virtual track.
According to another aspect of the present invention, there is also provided a system for generating a movement trajectory, including:
the data acquisition module is used for acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
the pre-training module is used for inputting the data of the historical track into a generator, predicting the next position point in the historical track through the generator and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
and the formal training module is used for repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and the track simulation module is used for generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
In accordance with at least one embodiment of the present invention, the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
According to at least one embodiment of the invention, the formal training module is further configured to:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
According to at least one embodiment of the present invention, the trajectory simulation module is further configured to generate a starting point of the simulated trajectory in the following manner:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
According to at least one embodiment of the present invention, the formal training module is further configured to generate, by using the generator obtained through pre-training, the remaining position points of the virtual trajectory in a random sampling manner after generating the partial position points of the virtual trajectory, so as to obtain the virtual trajectory.
An embodiment of the present invention further provides a computer-readable storage medium, where a computer program is stored on the computer-readable storage medium, and when the computer program is executed by a processor, the steps of the method for generating a movement trajectory as described above are implemented.
Compared with the prior art, the method and the device for generating the movement track provided by the embodiment of the invention at least have the following beneficial effects:
1) the complex characteristics of the moving track are extracted by using a recurrent neural network, and the complex transfer relation in the moving track can be modeled by combining various elaborately constructed moving behavior characteristics;
2) a track quality evaluation system is constructed based on a reinforcement learning and confrontation generation network and a movement rule is synthesized, and a generated model capable of generating a more real movement track can be obtained by automatically comparing the difference between the generated track and the real track to guide model training and updating.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments of the present invention will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained based on these drawings without inventive labor.
Fig. 1 is a system block diagram of a system for generating a movement trajectory according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a generator according to an embodiment of the present invention;
fig. 3 is a schematic flow chart of a method for generating a movement track according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a system for generating a movement trajectory according to an embodiment of the present invention;
fig. 5 is another schematic structural diagram of the movement trajectory generation system according to the embodiment of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments. In the following description, specific details such as specific configurations and components are provided only to help the full understanding of the embodiments of the present invention. Thus, it will be apparent to those skilled in the art that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the invention. In addition, descriptions of well-known functions and constructions are omitted for clarity and conciseness.
It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present invention. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
In various embodiments of the present invention, it should be understood that the sequence numbers of the following processes do not mean the execution sequence, and the execution sequence of each process should be determined by its function and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.
As described in the background art, the generation method of the movement trajectory in the prior art generally has the problems that it is difficult to simulate real and complex movement behaviors or reasonably evaluate simulation results. In order to solve at least one of the above problems, embodiments of the present invention provide a method for generating a movement trajectory, which uses a recurrent neural network to extract complex features of the movement trajectory, and combines various carefully constructed movement behavior features to model a complex transfer relationship in the movement trajectory. In addition, the embodiment of the invention also constructs a track quality evaluation system based on a reinforcement learning and confrontation generation network and a comprehensive movement rule, and guides the training and updating of the model by automatically comparing the difference between the generated track and the real track so as to obtain a generation model capable of generating a more real movement track.
Fig. 1 is a schematic diagram of a system block diagram of a system for generating a movement trajectory according to an embodiment of the present invention. The system 10 for generating a moving trajectory shown in fig. 1 employs a network architecture of a Generative Adaptive Network (GAN) including two main network models, i.e., a generator and a discriminator, and fig. 2 shows an example of a specific structure of a generator according to an embodiment of the present invention, where the generator includes an embedded layer, an LSTM layer, and a plurality of linear layers. And the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors. The LSTM layer is a bidirectional LSTM network and is used for receiving the feature vector sequence generated by the embedding layer and generating the vector representation of the next position point. The plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
Here, the embodiment of the present invention introduces a bidirectional LSTM network to perform timing feature extraction on unknown feature vectors, and the starting point is that the geographic location of the next time point (or next hop) of an individual is not only related to the previous time point of the individual, but also related to the subsequent time point of the next time point. For example, a student's decision regarding where to eat after a class is related to both the location of his class and the location of his class to be taken after a meal, so that the location of the next hop is not simply determined by the current location, and the above features can be captured more efficiently using the two-way LSTM network; the linear layer is to perform feature dimensionality reduction on the features in a multi-layer perception mode, generate probability distribution of the next hop at the full spatial position according to the feature dimensionality reduction, and usually take the position point with the maximum probability as the position of the next hop.
As shown in fig. 3, a method for generating a movement trajectory according to an embodiment of the present invention includes:
Here, the target area may be a city or other area with a certain spatial range, where trajectory simulation is required. The individuals may be pedestrians as well as various transportation vehicles (e.g., bicycles, automobiles, motorcycles, etc.). The embodiment of the invention acquires the pre-acquired data of the historical movement tracks of a plurality of individuals in the target area, wherein the data can generally comprise the geographic positions (such as longitude and latitude information) of the individuals and the time information when the geographic positions are acquired.
In addition, in order to reflect the movement law of the individual more, the data of the historical track may further include at least one of the following information: point of Interest (POI) information of an area where the geographical location is located, and context information of the geographical location. The contextual information of the geographic location is typically information related to the source of the trajectory data. For example, taking the historical track obtained from a social network as an example, the context of the geographic location may be a textual description of the geographic location; taking the historical track obtained from the operator of the mobile communication network as an example, the context of the geographic location may be related information of the base station that collected the geographic location, such as the location, coverage area, and number of sectors of the base station.
In step 32, embodiments of the present invention pre-train the generator and the arbiter. Before pre-training, data of historical tracks needs to be pre-processed, that is, each track extracts the geographical position and related information (such as POI and context) of an individual at preset time intervals, so as to obtain the geographical position and related information of multiple time points (also referred to as time points for short herein), the geographical position of each time point is taken as a position point (also referred to as track point for short herein) of the track, so as to obtain multiple position points, and attributes of each position point include time, geographical position, POI and context. The attributes of each location point are then converted into a high-dimensional feature vector (also referred to herein as a location point feature vector), such that each track can be characterized by a sequence of feature vectors for a plurality of location points of the track. And sequencing the feature vectors of each position point in the feature vector sequence according to the time sequence. For convenience of processing, the embodiment of the present invention may generate the feature vectors having a preset number of position points. For example, assuming that the sampled trajectory is the movement trajectory of an individual on a natural day, sampling may be performed every hour to generate a feature vector sequence of 24 location points.
Here, the data of the historical track is track data within a preset time length range, that is, the track data is a track recording an individual within a time range of a preset length, and the preset length may be 24 hours of a day, multiple days, or the like. Specifically, the embodiment of the invention can sample a preset number of position points from the historical track data of each individual movement, and preset time intervals are arranged between adjacent position points; then, dividing the target area into a plurality of grids, and determining the grid to which each position point belongs as the spatial attribute of the position point according to the geographic position of each position point; dividing the preset time range into a plurality of time intervals, and obtaining the time interval of each position point according to the time information of each position point as the time attribute of the position point; and then, generating a feature vector of each position point according to the spatial attribute and the time attribute of each position point to obtain a feature vector sequence of each historical track. Of course, when considering more attributes, such as POI and context information, the embodiment of the present invention may convert the attributes into corresponding feature vectors according to a word embedding algorithm, and form the feature vectors of the location points together with the geographic location and the time attributes.
In order to achieve faster convergence of the network model during the subsequent formal training process, the embodiment of the present invention pre-trains the generator and the arbiter in step 32. Through pre-training, the generator can learn some sequence transfer rules, and the discriminator can distinguish a real track from a disturbance track. Here, the perturbation track may be obtained by perturbing a part of the location points in the real historical track, for example, changing the geographic location of a certain location point or certain location points in the historical track, and/or changing the time of a certain location point in the historical track, and so on.
After the pre-training in step 32, the embodiment of the present invention performs a formal training on the network model through step 33. In the formal training process, training is carried out based on the principle of a generative confrontation network. Wherein the generator generates a virtual trajectory and provides it to the arbiter. The discriminator discriminates the virtual track and the real historical track to obtain a discrimination result whether the virtual track is the real track, and a reinforcement learning algorithm is used for generating a gradient corresponding to the discrimination result and feeding the gradient back to the generator, so that the generator updates the model parameters until a preset convergence condition is reached, for example, the discriminator cannot distinguish the real historical track from the virtual track generated by the generator. In practical applications, the convergence condition may be set as: the success rate of distinguishing the virtual track by the discriminator is less than a preset threshold.
In addition, when the generator generates the virtual track, the generator may determine a starting point of the virtual track, generate a position point of a next hop based on the starting point, combine the position point with the generated position point, and continue to generate subsequent position points until the virtual track including the preset number of position points is generated. The starting point of the virtual trajectory may be generated based on population density of the target area according to population distribution characteristics of the target area, for example, population density of the population in the target area at a preset starting time period (for example, a time period when the individual movement trajectory starts mostly in the early morning or in the morning), wherein the probability that a certain location point is selected as the starting point is higher as the population density of the location point is higher.
And step 34, generating a simulation track of the target area by using the generator according to the starting point of the simulation track.
Through steps 31-33, a generator can be obtained through training, and the generator can be used for generating a subsequent simulation track. When the simulation track is generated, according to the data of the historical track of the target area, the embodiment of the invention can count the population distribution characteristics in the preset initial period, such as population density in each grid; determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid; then, the starting point of the simulation track is generated according to the distribution probability of the starting point in the target area. And generating the position point of the next jump hop by utilizing the generator hop by hop based on the starting point to obtain one or more complete simulation tracks.
Through the steps, the embodiment of the invention realizes the generation of the simulation track. In the above process, the embodiment of the present invention uses the recurrent neural network to extract the complex features of the movement trajectory, and models the complex transfer relationship in the movement trajectory. In addition, a track quality evaluation system is constructed based on a reinforcement learning and confrontation generation network and a movement rule is synthesized, and a generation model capable of generating a more real movement track can be obtained by guiding model training and updating through automatically comparing the difference between the generation track and the real track.
According to at least one embodiment of the present invention, in the step 33, when the discriminator determines the track authenticity, the embodiment of the present invention further introduces more rules of individual mobility, such as distance rules and periodic rules, for example, the individual movement is affected by the traffic vehicle and the traffic conditions that can be provided by the target area, and the movement distance is usually limited, which can be reflected by the distance rules of the individual movement.
For another example, the movement of an individual is typically periodic, e.g., the individual is typically located at a company during work hours on different weekdays, at home in the morning or evening of a weekday, i.e., the individual is typically located at the same geographic location or a relatively small distance between geographic locations at the same point in time on different natural days, and at some point in time on the same natural day. Therefore, the embodiment of the invention can preset related time points, including the same periodic time point and different time points set according to the periodicity of the individual movement law, and reflect the periodic law of the individual movement through the geographic position of the related time points.
Based on the above analysis, when the virtual trajectory and the sampled historical trajectory are determined in step 33 and a determination result indicating whether the virtual trajectory is real is obtained, the embodiment of the present invention may include:
1) distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
2) calculating the sum of the distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum, and whether the virtual track conforms to the distance rule of the movement of the individual can be reflected;
3) calculating the difference of the geographic positions between the relevant time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively correlated with the difference, so that whether the virtual track conforms to the periodic rule of the movement of the individual can be reflected; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
4) and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result. For example, the first, second, and third determination results may be weighted and summed to obtain a final determination result, which is used to determine whether the virtual trajectory is real, where whether the virtual trajectory is real refers to whether the virtual trajectory is similar to the real trajectory in terms of characteristic morphology.
In the case where the discriminator performs discrimination, the discriminator may discriminate between a plurality of virtual trajectories and generate a final discrimination result from the discrimination results of these virtual trajectories.
Through the processing, the embodiment of the invention introduces more characteristics capable of reflecting the real moving track rule into the discriminator, so that the discrimination performance of the virtual track can be improved, the updating of the generator is guided, and the generator with better performance is obtained.
In the process that the generator generates the preset number of position points of the virtual sequence, a large offset from the real track may be generated from a certain position point, and on the basis, more position points of the track generated subsequently are far away from the real track, so that the finally generated virtual track has a large morphological difference from the real track. According to at least one embodiment of the present invention, in order to improve the model training efficiency, in step 33, after the generator obtained by pre-training is used to generate partial position points of the virtual trajectory, the embodiment of the present invention may generate the remaining position points of the virtual trajectory in a random sampling manner, so as to obtain the virtual trajectory. Here, each position point may be represented by a feature vector corresponding to the position point, and each trajectory may be represented by a feature vector sequence composed of feature vectors of all position points of the trajectory. In addition, the number of the partial position points may be set in advance. For example, assuming that the number of position points of each trajectory is 24, it may be set to 5, 8, or 12, and so on.
That is, in the countermeasure training, the generator samples and acquires the specific position of the next hop according to the next hop visit probability, and combines the position and the historical hop position into a partial sequence. Because the discriminator can not accept the partial sequence as input, the partial sequence needs to firstly use the generator trained at present to carry out a completion operation, namely the partial sequence is input into the generator and samples each jump in the same way until the completion is completed; because the sampling process has randomness, and an independent sampling process cannot fairly give out reward (reward) values, the embodiment of the invention can Search in the space formed by all completion data through a Monte Carlo Tree Search (MC) algorithm, inputs the Search result into a discriminator to discriminate, obtains the probability average value output by the discriminator as the reward value, and updates the generation model responsible for the completion sequence and the generation model responsible for generating the jump according to the reward gradient corresponding to the strategy; the updating of the main generation model (i.e. the model responsible for the jump) is to take forwards as loss (loss) to perform backward propagation on the network and update the weight according to the gradient, and then to delay the updating of the completion model on the updated network according to a certain rate, because the task of the completion model is different from that of the jump model, the jump model cannot be followed blindly, otherwise, it is not meaningful to design such a completion mechanism; and then, the generator is used for generating a batch of data of the virtual track, the data of the virtual track and the data of the real track are mixed and input into the discriminator, and the identification capability of the discriminator is trained.
Based on the above method for generating the movement track, the embodiment of the invention also provides a device for implementing the method.
Referring to fig. 4, a system 40 for generating a movement track according to an embodiment of the present invention includes:
the data acquisition module 41 is configured to acquire pre-acquired data of historical trajectories of movement of multiple individuals in a target area, where the data of the historical trajectories at least include geographic positions and time information of the individuals;
the pre-training module 42 is used for inputting the data of the historical track into a generator, predicting the next position point in the historical track through the generator and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
a formal training module 43, configured to repeatedly perform the following iterative processes until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and a track simulation module 44, configured to generate, by using the generator, a simulation track of the target area according to a starting point of the simulation track.
Through the modules, the embodiment of the invention can simulate real auxiliary movement behaviors, and guides the training and updating of the model by comparing the generated track with the real track automatically, so that the model capable of generating a more real movement track can be obtained.
In accordance with at least one embodiment of the present invention, the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
According to at least one embodiment of the invention, the data of the historical track is track data within a preset time length range. Specifically, the embedding layer samples a preset number of position points from data of a historical track of movement of each individual, and a preset time interval is arranged between adjacent position points; dividing the target area into a plurality of grids, and determining the grid to which each position point belongs according to the geographic position of each position point as the spatial attribute of the position point; dividing the preset time range into a plurality of time intervals, and obtaining the time interval of each position point according to the time information of each position point as the time attribute of the position point; and generating a feature vector of each position point according to the spatial attribute and the time attribute of each position point to obtain a feature vector sequence of each historical track.
According to at least one embodiment of the invention, the data of the historical track further comprises at least one of the following information: POI information of interest points in the area where the geographic position is located, and context information of the geographic position.
According to at least one embodiment of the present invention, the preset convergence condition is: the success rate of distinguishing the virtual track by the discriminator is less than a preset threshold.
According to at least one embodiment of the present invention, the formal training module 43 is further configured to:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
According to at least one embodiment of the present invention, the trajectory simulation module is further configured to generate a starting point of the simulated trajectory in the following manner:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
According to at least one embodiment of the present invention, the formal training module is further configured to generate, by using the generator obtained through pre-training, the remaining position points of the virtual trajectory in a random sampling manner after generating the partial position points of the virtual trajectory, so as to obtain the virtual trajectory.
As shown in fig. 5, another moving trace generating system 50 is further provided in the embodiment of the present invention, where the moving trace generating system 50 specifically includes a processor 51, a memory 52, a bus system 53, a receiver 54, and a transmitter 55. Wherein, the processor 51, the memory 52, the receiver 54 and the transmitter 55 are connected through the bus system 53, the memory 52 is used for storing instructions, the processor 51 is used for executing the instructions stored in the memory 52 to control the receiver 54 to receive signals and control the transmitter 55 to transmit signals;
the processor 51 is configured to read a program in the memory, and execute the following processes:
acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
inputting the data of the historical track into a generator, predicting a next position point in the historical track through the generator, and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
It should be understood that, in the embodiment of the present invention, the processor 51 may be a Central Processing Unit (CPU), and the processor 51 may also be other general processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field Programmable Gate Arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the like. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like.
The memory 52 may include a read-only memory and a random access memory, and provides instructions and data to the processor 51. A portion of the memory 52 may also include non-volatile random access memory. For example, the memory 52 may also store device type information.
The bus system 53 may include a power bus, a control bus, a status signal bus, and the like, in addition to the data bus. For clarity of illustration, however, the various buses are labeled in the figure as bus system 53.
In implementation, the steps of the above method may be performed by integrated logic circuits of hardware or instructions in the form of software in the processor 51. The steps of a method disclosed in connection with the embodiments of the present invention may be directly implemented by a hardware processor, or may be implemented by a combination of hardware and software modules in the processor. The software module may be located in ram, flash memory, rom, prom, or eprom, registers, etc. storage media as is well known in the art. The storage medium is located in the memory 52, and the processor 51 reads the information in the memory 52 and completes the steps of the method in combination with the hardware thereof. To avoid repetition, it is not described in detail here.
In accordance with at least one embodiment of the present invention, the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
According to at least one embodiment of the invention, the data of the historical track is track data within a preset time length range; the program when executed by the processor 51 may further implement the steps of:
sampling a preset number of position points from data of a historical track of movement of each individual, wherein a preset time length is arranged between every two adjacent position points;
dividing the target area into a plurality of grids, and determining the grid to which each position point belongs according to the geographic position of each position point as the spatial attribute of the position point; dividing the preset time range into a plurality of time intervals, and obtaining the time interval of each position point according to the time information of each position point as the time attribute of the position point;
and generating a feature vector of each position point according to the spatial attribute and the time attribute of each position point to obtain a feature vector sequence of each historical track.
According to at least one embodiment of the invention, the data of the historical track further comprises at least one of the following information: POI information of interest points in the area where the geographic position is located, and context information of the geographic position.
According to at least one embodiment of the present invention, the preset convergence condition is: the success rate of distinguishing the virtual track by the discriminator is less than a preset threshold.
According to at least one embodiment of the invention, the program when executed by the processor 51 may further implement the steps of:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
According to at least one embodiment of the invention, the program when executed by the processor 51 may further implement the steps of:
generating a starting point of the simulation track according to the following modes:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
According to at least one embodiment of the invention, the program when executed by the processor 51 may further implement the steps of:
and generating the rest position points of the virtual track in a random sampling mode after generating part position points of the virtual track by using the generator obtained by pre-training to obtain the virtual track.
In some embodiments of the invention, there is also provided a computer readable storage medium having a program stored thereon, which when executed by a processor, performs the steps of:
acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
inputting the data of the historical track into a generator, predicting a next position point in the historical track through the generator, and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
When being executed by the processor, the program can implement all implementation manners in the method for generating a movement trajectory shown in fig. 3, and can achieve the same technical effect, and is not described herein again to avoid repetition.
Those of ordinary skill in the art will appreciate that the various illustrative elements and algorithm steps described in connection with the embodiments disclosed herein may be implemented as electronic hardware or combinations of computer software and electronic hardware. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the implementation. Skilled artisans may implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present invention.
It is clear to those skilled in the art that, for convenience and brevity of description, the specific working processes of the above-described systems, apparatuses and units may refer to the corresponding processes in the foregoing method embodiments, and are not described herein again.
In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other ways. For example, the above-described apparatus embodiments are merely illustrative, and for example, the division of the units is only one logical division, and other divisions may be realized in practice, for example, a plurality of units or components may be combined or integrated into another system, or some features may be omitted, or not executed. In addition, the shown or discussed mutual coupling or direct coupling or communication connection may be an indirect coupling or communication connection through some interfaces, devices or units, and may be in an electrical, mechanical or other form.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment of the present invention.
In addition, functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may exist alone physically, or two or more units are integrated into one unit.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (13)
1. A method for generating a movement trajectory is characterized by comprising the following steps:
acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
inputting the data of the historical track into a generator, predicting a next position point in the historical track through the generator, and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
2. The method of claim 1, wherein the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
3. The method of claim 2, wherein the data of the historical track is track data within a preset time length range; generating a feature vector sequence for characterizing each historical track according to the historical track data, wherein the step comprises the following steps:
sampling a preset number of position points from data of a historical track of movement of each individual, wherein a preset time length is arranged between every two adjacent position points;
dividing the target area into a plurality of grids, and determining the grid to which each position point belongs according to the geographic position of each position point as the spatial attribute of the position point; dividing the preset time range into a plurality of time intervals, and obtaining the time interval of each position point according to the time information of each position point as the time attribute of the position point;
and generating a feature vector of each position point according to the spatial attribute and the time attribute of each position point to obtain a feature vector sequence of each historical track.
4. The method of claim 1, wherein the data of the historical track further comprises at least one of: POI information of interest points in the area where the geographic position is located, and context information of the geographic position.
5. The method of claim 1, wherein the preset convergence condition is: the success rate of distinguishing the virtual track by the discriminator is less than a preset threshold.
6. The method of claim 1, wherein the step of obtaining the discrimination result by discriminating the virtual trajectory and the sampled historical trajectory using the discriminator comprises:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
7. The method of claim 1, wherein the starting point of the simulation trajectory is generated by:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
8. The method of any one of claims 1 to 7, wherein the step of generating a virtual trajectory using the pre-trained generator comprises:
and generating the rest position points of the virtual track in a random sampling mode after generating part position points of the virtual track by using the generator obtained by pre-training to obtain the virtual track.
9. A system for generating a movement trajectory, comprising:
the data acquisition module is used for acquiring pre-acquired data of historical tracks of the movement of a plurality of individuals in a target area, wherein the data of the historical tracks at least comprise the geographical positions and time information of the individuals;
the pre-training module is used for inputting the data of the historical track into a generator, predicting the next position point in the historical track through the generator and pre-training the generator; inputting the historical track and the disturbance track into a discriminator, distinguishing the disturbance track by training the discriminator, and pre-training the discriminator, wherein the disturbance track is obtained by performing disturbance processing on part of position points in the historical track;
and the formal training module is used for repeatedly executing the following iterative processing until a preset convergence condition is reached: generating a virtual track by using the generator obtained by pre-training; inputting the virtual track to the discriminator obtained by pre-training, and discriminating the virtual track and the historical track obtained by sampling by using the discriminator to obtain a discrimination result; generating a gradient corresponding to the judgment result by using a reinforcement learning algorithm, reversely transmitting the gradient to the generator, and updating the generator;
and the track simulation module is used for generating the simulation track of the target area by using the generator according to the starting point of the simulation track.
10. The apparatus of claim 9, wherein the generator comprises an embedding layer, a bi-directional LSTM network, and a plurality of linear layers, wherein,
the embedding layer generates a feature vector sequence for representing each historical track according to the data of the historical tracks, wherein the feature vector sequence comprises a preset number of position point feature vectors;
the bidirectional LSTM network receives the feature vector sequence generated by the embedding layer and generates a vector representation of a next position point;
the plurality of linear layers map the vector representation of the next location obtained by the bidirectional LSTM network to the geographic location of the next location point.
11. The apparatus of claim 9, wherein the formal training module is further configured to:
distinguishing the virtual track and the sampled historical track by using the discriminator to obtain a first judgment result of the authenticity of the virtual track;
calculating a sum of distances between all adjacent position points on the virtual track, and generating a second judgment result of the authenticity of the virtual track according to the sum, wherein the authenticity is negatively related to the sum;
calculating the difference of the geographic positions between the related time points on the virtual track, and generating a third judgment result of the authenticity of the virtual track according to the difference, wherein the authenticity is negatively related to the difference; the related time points comprise the same periodic time point and different time points which are set according to the periodicity of the individual movement rule;
and generating a judgment result of the authenticity of the virtual track according to the first judgment result, the second judgment result and the third judgment result.
12. The apparatus of claim 9,
the track simulation module is further configured to generate a starting point of the simulation track in the following manner:
counting population distribution characteristics at a preset initial period according to the data of the historical track of the target area;
determining the distribution probability of the starting point of the simulation track in the target area according to the population distribution characteristics, wherein the distribution probability of the starting point in one grid is positively correlated with the population density in the grid;
and generating a starting point of the simulation track according to the distribution probability of the starting point in the target area.
13. The apparatus of any of claims 9 to 12,
the formal training module is further configured to generate, by using the generator obtained through pre-training, a part of position points of the virtual trajectory, and then generate, by using a random sampling manner, remaining position points of the virtual trajectory to obtain the virtual trajectory.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010086899.9A CN113255951A (en) | 2020-02-11 | 2020-02-11 | Method and device for generating movement track |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010086899.9A CN113255951A (en) | 2020-02-11 | 2020-02-11 | Method and device for generating movement track |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113255951A true CN113255951A (en) | 2021-08-13 |
Family
ID=77219586
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010086899.9A Pending CN113255951A (en) | 2020-02-11 | 2020-02-11 | Method and device for generating movement track |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113255951A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115601393A (en) * | 2022-09-29 | 2023-01-13 | 清华大学（Cn） | Track generation method, device, equipment and storage medium |
-
2020
- 2020-02-11 CN CN202010086899.9A patent/CN113255951A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115601393A (en) * | 2022-09-29 | 2023-01-13 | 清华大学（Cn） | Track generation method, device, equipment and storage medium |
| CN115601393B (en) * | 2022-09-29 | 2024-05-07 | 清华大学 | Track generation method, track generation device, track generation equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Xu et al. | Real-time prediction of taxi demand using recurrent neural networks | |
| Liu et al. | Urban traffic prediction from mobility data using deep learning | |
| CN115240425B (en) | Traffic prediction method based on multi-scale space-time fusion graph network | |
| CN111653088B (en) | Vehicle driving quantity prediction model construction method, prediction method and system | |
| CN107679557B (en) | Driving model training method, driver identification method, device, equipment and medium | |
| CN110414732B (en) | Travel future trajectory prediction method and device, storage medium and electronic equipment | |
| CN109142171A (en) | The city PM10 concentration prediction method of fused neural network based on feature expansion | |
| CN109920248B (en) | Bus arrival time prediction method based on GRU neural network | |
| CN109544690A (en) | Shared bicycle trip influence factor recognition methods, system and storage medium | |
| CN110944295B (en) | Position prediction method, position prediction device, storage medium and terminal | |
| CN113780665B (en) | Private car stay position prediction method and system based on enhanced recurrent neural network | |
| CN110795522B (en) | Method and device for predicting track position of mobile user | |
| Chen et al. | ASC: Actuation system for city-wide crowdsensing with ride-sharing vehicular platform | |
| Cheng et al. | Developing a travel time estimation method of freeway based on floating car using random forests | |
| Zhang et al. | CGAIL: Conditional generative adversarial imitation learning—An application in taxi Drivers’ strategy learning | |
| CN116108984A (en) | Urban flow prediction method based on flow-POI causal relationship reasoning | |
| CN113222239A (en) | Short-time traffic flow prediction method based on CNN-LSTM-At neural network | |
| CN114461931A (en) | User trajectory prediction method and system based on multi-relation fusion analysis | |
| CN115099328A (en) | Traffic flow prediction method, system, device and storage medium based on countermeasure network | |
| Xu et al. | Simulating continuous-time human mobility trajectories | |
| CN117037483A (en) | Traffic flow prediction method based on multi-head attention mechanism | |
| CN113255951A (en) | Method and device for generating movement track | |
| CN113326877A (en) | Model training method, data processing method, device, apparatus, storage medium, and program | |
| CN113159371A (en) | Unknown target feature modeling and demand prediction method based on cross-modal data fusion | |
| CN106779181A (en) | Method is recommended by a kind of medical institutions based on linear regression factor Non-negative Matrix Factorization model |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |